Human Septic Myopathy: Induction of Cyclooxygenase, Heme Oxygenase and Activation of the Ubiquitin Proteolytic Pathway

Two groups of patients were included in the study. One group (table 1) (n = 24, septic group) consisted of patients with sepsis as defined by Bone et al.  23and at least one of the following signs of organ dysfunction: systolic blood pressure <90 mm Hg despite adequate fluid replacement or need for vasopressors to maintain blood pressure; oliguria (urine output <0.5 ml·kg⁻¹·h⁻¹); metabolic acidosis (pH <7.3 or lactate >2 mmol/l); Pao²/Fio²ratio <200 mm Hg; platelet count <100,000/mm³; or altered mental status. The other group (n = 10, control group) consisted of patients undergoing elective laparotomy or hip replacement for nonseptic conditions. Patients with cancer, chronic inflammatory processes, chronic heart failure or patient receiving glucocorticoids drugs were excluded from study as these may alter expression of nitric oxide synthase, COX, and HO proteins.24,25The duration of immobilization was defined as the time during which patients were kept in bed until the day of biopsy.

To assess the severity of the clinical conditions, the Simplified Acute Physiology Score II26and the Logistic Organ Dysfunction Score27,28were calculated for each patient at the time of biopsy. In living patients, muscle weakness was evaluated when the patient recovered from sepsis and was defined as decrease of muscular force.

The study was performed with the approval of the local boards governing research on human subjects at our institution (Comité Consultatif pour la Protection des Personnes dans la Recherche Biomédicale, Hôpital Saint-Louis, Paris, France, and Ethics Committee, University College of London Hospitals, London, UK). All patients or theirs relatives gave informed written consent.

A biopsy specimen was obtained from either the rectus abdominis or vastus lateralis muscle within 24 h after the onset of sepsis and mechanical ventilation. The former was obtained during the initial phase of the laparotomy: after skin incision and dissection through the subcutaneous fat, the anterior sheet of the rectus muscle was opened with scissors and a biopsy specimen obtained. Sampling was not performed if the muscle was in direct contact with the site of infection, thus avoiding tissue where protein expression may be directly affected by local inflammation. Biopsies from vastus lateralis were taken percutaneously from septic patients in the intensive care unit, using Henckel Tilley forceps via  a small incision through the skin with blunt dissection through subcutaneous tissue and muscle capsule.18For control group, the biopsies were obtained during the initial phase of either laparotomy (rectus abdominis) or hip surgery (vastus lateralis). Both control and septic patients received paralytic agents only during surgery.

Tissue samples were immediately frozen into liquid nitrogen and stored at −80°C for <6 months before subsequent measurement of protein expression. The various assays and the reading of immunostained biopsies were performed in a blinded fashion.

Tissue samples were homogenized in a Triton lysis buffer. Proteins in the tissue homogenates were separated by electrophoresis on 7.5% sodium dodecyl sulfate polyacrylamide gel in a electrophoretic buffer. They were transferred overnight at 4°C to a nitrocellulose membrane (Bio-Rad Laboratories, Marnes la Coquette, France). Membranes were first blocked at room temperature with 5% nonfat dry milk, bovine serum albumin 1% in a Tris-buffered saline Tween; they were then incubated with a COX-1 monoclonal antibody (Oxford Biomedical Research/Euromedex, Mundolsheim, France) at a 1:1000 dilution for COX-1 protein and with a COX-2 monoclonal antibody (Oxford Biomedical Research/Euromedex) at a 1:200 dilution for COX-2. A lysate of human umbilical vascular endothelial cells stimulated with 4-Phorbol-12-Myristate-13-Acetate was used as a positive control for detection of COX-2 protein. Antibody detection was performed with a chemiluminescence substrate using the Amersham ECL kit (Amersham Biosciences, Freiburg Germany). After detection, membranes were stained with Red Ponceau solution and the total amount of proteins was assessed. Protein expression was analyzed with a Las 1000-plus system (Raytest, Fugifilm, St Quentin en Yvelines, France) and protein quantification was obtained with Image Gauge Ver 3,4 software (Fugifilm).

Muscle samples obtained from rectus abdominis were frozen quickly in liquid nitrogen and embedded in paraffin. Five-micron sections were mounted on poly-L-lysine-coated microscope slides. The sections were rinsed in phosphate-buffered saline and incubated in blocking buffer. Slides were then incubated with a COX-2 rabbit polyclonal antibody (L320 R6, kindly donated by J. Maclouf, Ph.D. and H. Habib, Ph.D., Inserm U.348, Hopital Lariboisière, Paris, France) and a COX-1 rabbit polyclonal antibody (L6 R3, donated by J. Maclouf and H. Habib) at the antibody concentration of 1:100 for both proteins. After washing in phosphate-buffered saline, slides were exposed to a biotinylated antirabbit 5% human serum (BioRad) diluted to 1:100. Localization of COX-1 and COX-2 was revealed with streptavidin fluoresceine rabbit antibody (BioRad) at 1:50 dilution. The fluorescence of the slides was examined by a microscope Leica. Specificity of immunostaining was evaluated by omission of the primary antibody and by using pooled nonimmune antirabbit immunoglobulin G instead of the primary antibody at equivalent concentration, and processed as above.

Proteins were separated by electrophoresis on 12% sodium dodecyl sulfate polyacrylamide gel. The membranes were incubated with a mouse HO-1 monoclonal antibody (Stressgen/Tebu SA, Le Perray en Yvelines, France) used at a 1:2000 dilution for HO-1 protein and with a rabbit HO-2 polyclonal antibody (Stressgen/Tebu SA) used at a 1:1000 dilution for HO-2. HO-1 peptide SPT-896 (Stressgen/Tebu SA) was used as a positive control for HO-1.

Ubiquitin activity was performed by Western blot in tissue samples homogenized in sodium dodecyl sulfate instead of Triton. Proteins were separated by electrophoresis on 12% sodium dodecyl sulfate polyacrylamide gel and membranes were incubated with a mouse ubiquitin monoclonal antibody (Santa-Cruz Biotechnologies/Tebu SA) used at a 1:100 dilution. This allowed quantification and comparison of the expression of free ubiquitin protein in tissue samples of control and septic patients. The positive control of free ubiquitin protein was full length ubiquitin of human origin produced in Escherichia Coli  as 35 kD when tagged to its fusion protein (Santa Cruz Biotechnology/Tebu SA).

When ubiquitin is active, it binds to altered proteins to initiate their breakdown. To evaluate the activity of ubiquitin pathway in our samples, we quantified the concentration of ubiquitinated proteins (as previously described)29,30by Western blot. Values were standardized with the Ponceau S intensity.

Sodium dodecyl sulfate, glycerol, 2-mercaptoethanol, bromophenol blue, and the electrophoretic system were obtained from Bio-Rad (Marnes la Coquette, France). Chemiluminescence substrates were obtained from Amersham Biosciences (Freiburg, Germany). All other chemicals were purchased from Sigma-Aldrich (St Quentin Fallavier, France).

Values are given as median (IQR) except age, Simplified Acute Physiology Score II, and Logistic Organ Dysfunction Score (mean ± SD). All results obtained from rectus abdominis and vastus lateralis biopsies were similar in control patients and are thus pooled. Differences were analyzed by the Mann-Whitney U -test. The relation between free ubiquitin protein expression and ubiquitinated proteins expression was assessed by linear regression analysis. Statistical significance was accepted at P < 0.05.

Characteristics of the septic patients are given in table 1. Age was similar between the control (n = 10) and septic (n = 24) groups (mean ± SD, 54 ± 12 yr versus  64 ± 19 yr, respectively P = 0.13). Organ injury was reflected by a higher Simplified Acute Physiology Score II in septic compared with control patients (59 ± 18 versus  11 ± 4, P < 0.0001), by a high Logistic Organ Dysfunction Score (7 ± 3), and by a prolonged duration of ventilatory support (13 ± 14 days) in the septic group. Eleven patients died within 14 days in septic group; none died in the control group. The median value of the length of immobilization in septic patients was 0 days (extreme values, 0–18 days). Four of the 13 septic patients who survived developed muscle weakness.

Muscular homogenates expressed more COX-2 protein, the inducible isoform of COX, as detected by Western blot, in septic compared with control patients (fig. 1A). Quantification of COX-2 protein expression confirmed the Western blot results and showed a median value of 13.4 (3.5–39.8) arbitrary units in septic patients versus  4.0 (0–17.7) arbitrary units in control patients (P < 0.005 versus  control, fig. 1B). Figure 1Calso shows that expression of COX-2 in muscle is higher in patients that will die (16.2 [6.1–36.1] arbitrary units) versus  those who will survive (6.3 [0.0–39.8] arbitrary units, P < 0.006). No difference in COX-1 expression was observed between septic and control groups; the median values were 20.8 (0–63.0) arbitrary units and 20.0 (9.3–48.8) arbitrary units, respectively (P = 0.44).

Histologic analysis showed no inflammatory infiltrate and an absence of tissue edema in the muscles of both septic and control patients (fig. 2). It also showed increased COX-2 staining, reproducibly observed in the sarcolemma of skeletal muscle myocytes of all septic patients expressing COX-2 protein as shown by Western blot. By contrast, a very low level of COX-2 staining was found in control patients. Figure 2also shows similar COX-1 staining in the control and septic patients, mostly localized to the sarcolemma.

Western Blot analysis (fig. 3A) revealed that the expression of inducible HO-1 protein was greater in muscle homogenates taken from septic patients compared with that taken from control patients. This was confirmed by quantification that showed a median value of HO-1 expression of 13.4 (0–60.0) arbitrary units in rectus abdominis muscle of septic patients versus  0.4 (0–18.0) arbitrary units in control patients (P < 0.01, fig. 3B). Figure 3Calso shows that expression of HO-1 in muscle is higher in patients that will die (22.4 [8.7–60.0] arbitrary units) versus  those who will survive (3.9 [0.0–19.9] arbitrary units) (P < 0.006). Patients who did not develop muscle weakness tended to express more HO-1 than patients who develop muscle weakness: 12.0 (1.0–1.8) arbitrary units and 1.1 (0–2.2) arbitrary units (P = 0.07), respectively. No difference in HO-2 expression was detected between control and septic groups (data not shown).

Higher concentrations of both free ubiquitin and ubiquitinated proteins were detected in muscle homogenates of septic patients compared with control patients (figs. 4 and 5). Quantification of free ubiquitin protein expression confirmed Western blot results and showed a median value of 10.5 (6.1–20.6) arbitrary units in septic patients versus  6.9 (2.7–9.7) arbitrary units in muscle biopsies of control patients (P < 0.02, fig. 4B). This supports a higher ubiquitin proteolytic activity that affected several proteins in human sepsis. Indeed, figure 5Ashows that ubiquitinated proteins are more marked in septic than in control patients with several bands having molecular mass of 160, 130, and 60 kD, which were consistently detected in septic patients but lacking in control patients. By contrast, one ubiquitinated protein with an apparent molecular mass of 180 kD was present in control patients but was not found in septic patients. No difference in ubiquitin activity was detected between muscle homogenates from either thigh or abdominal muscles (data not shown). Quantification of all muscle ubiquitinated proteins showed that the concentration was greater in septic compared with control patients: 92.8 (26.5–128.9) arbitrary units versus  49.8 (32.8–79.9) arbitrary units, respectively (P < 0.02, fig. 5B) and that the concentration of ubiquitinated proteins paralleled the expression of ubiquitin protein (r = 0.60, P < 0.003) (fig. 5C).

Expression of free ubiquitin and ubiquitinated proteins were similar between patients immobilized >24 h compared to those <24 h (data not shown) and between patients that died <14 days versus  patients that survived.

The current study shows the overexpression of two other major inducible enzymes involved in the inflammatory cascade (the proinflammatory COX-2 and the antiinflammatory HO-1) and stimulation of an active proteolytic ubiquitin-proteasome pathway, in anatomically different muscles of septic patients.

The immune reaction to microbial invasion is increasingly seen as a local proinflammatory response limited by protective systemic antiinflammatory responses.31,32Sepsis-induced nitric oxide synthase overexpression has been repeatedly described as a major player in the proinflammatory reaction to microbial invasion in mammals. Indeed, we and others have previously identified overexpression and increased activity of nitric oxide synthase in septic patients in various tissues including vascular wall,33left ventricular wall,34and skeletal muscle.5The current study further shows overexpression of another highly potent proinflammatory protein, COX-2, in muscle biopsies of rectus abdominis and vastus lateralis of septic, but not control, patients. COX-2 was mostly located in sarcolemma. COX-2 end products have been implicated in the skeletal muscle12,35–37and myocardial contractile failure in septic animals.38Accordingly, our findings strongly suggest a role of COX-2 in human sepsis although its specific effects on the inflammatory cascade and skeletal muscle contractile dysfunction remain to be explored.

In addition to these proinflammatory pathways, we also demonstrate the existence of a tissue-based antiinflammatory pathway, namely overexpression of HO-1, in our septic patients. Skeletal muscle overexpression of HO-1 that is accompanied by increased HO activity has been shown to play a protective role in the muscle of lipopolysaccharide-treated rats.13Our study suggests that HO-1 could have a protective role in skeletal muscle taken from septic patients because HO-1 expression tended to be greater in patients who did not develop muscle weakness. Although the exact role of HO-1 remains to be explored, one may suggest that it limits the local consequences of the proinflammatory effects of nitric oxide synthase and COX-2, including peroxynitrite production and protein nitration/inactivation.

We recently showed that muscle failure is at least partly related to nitric oxide synthase–induced peroxynitrite production in septic patients.5Peroxynitrite, a powerful oxidant, nitrates tyrosine residues into 3-nitrotyrosine, which likely alters protein function.17It is, however, unknown whether those nitrated proteins are degraded and what pathway is involved in protein breakdown in sepsis. More globally, mechanisms of sepsis-related muscle wasting are unknown10and might be related to a stimulation of catabolic pathways. Of these pathways, the ubiquitin-proteasome pathway39was described to be more specifically implicated in the breakdown of abnormal proteins, including peroxynitrite-induced nitrated proteins.21,22Indeed SIN-1, a peroxynitrite donor, has been shown to stimulate the proteasome pathway, leading to an increased rate of protein degradation, including myosin degradation.21Furthermore, inhibition of the proteasome pathway leads to an increased cellular content of 3-nitrotyrosine and ubiquitinated proteins.40In septic mammals, ubiquitin-proteasome activity is increased in rat skeletal muscle,19,41,42whereas both ubiquitin mRNA and myofibrillar proteolysis are increased in septic patients.20Our current study extends these previous studies by showing, for the first time, increased expression of an active ubiquitin protein in the skeletal muscles of septic patients. Indeed, our study shows a parallel increase in free ubiquitin expression and in ubiquitinated proteins in various muscles of septic compared with control patients. This suggests that the newly expressed ubiquitin protein binds to various “abnormal” proteins with different molecular mass (the ubiquitinated proteins) to activate the proteasome proteolytic pathway leading to muscle wasting. Detection of ubiquitinated proteins suggests that ubiquitin-conjugating enzymes and ligases42–44are also active in human septic muscle. Detection of ubiquitinated proteins could indicate an accumulation of proteins, suggesting an inhibition of proteasome activity. Studies have shown that expression of proteasome components are increased and that its proteolytic activity is also increased during sepsis in muscle.41We found that expression of free ubiquitin and ubiquitinated proteins, in the early phase of sepsis, was similar between patients who develop muscle weakness and patients who do not. Although we did not measure a marker of myofibrillar catabolism, this suggests that muscle catabolism is one of the pathways leading to muscle weakness in septic patients.

Others factors, such as immobilization, are known to be involved in the stimulation of the ubiquitin proteolytic pathway.41Nevertheless, in our study, this factor does not seem to influence the ubiquitin pathway because no difference in the expression of free ubiquitin and ubiquitinated proteins was found between immobilized and nonimmobilized septic patients.

Proinflammatory pathway activation and its consequences on muscle function have been primarily studied in the rectus abdominis in septic patients.5,16However, these biopsies, taken from patients with peritonitis or postcardiac surgical mediastinitis, may have been contaminated by the infected area. The current study extends previous work by showing that COX-2, HO-1, and ubiquitin activity are similarly overexpressed in rectus abdominis and in muscles distant from the infected area, namely the vastus lateralis. Accordingly, our data suggest that the protein overexpression and contractile failure previously observed in abdominal wall muscle5is a generalized phenomenon in septic patients and is related to circulating diffusible factors rather than local factors.

In summary, our study demonstrated a marked involvement of local proinflammatory and antiinflammatory pathways and of an active ubiquitin proteolytic pathway in skeletal muscle of septic patients. Involvement of ubiquitin pathway in other sepsis-induced organ dysfunction, including myocardial dysfunction, remains to be investigated in septic patients.

The authors would like to acknowledge Jane-Lise Samuel, M.D., Ph.D. (Researcher, IndianaSERM U572, Paris, France) for helpful discussion of the manuscript and to Joaquim Matéo, M.D., Florence Riché, M.D., Marie-Jo Laisné, M.D. (Consultants, Department of Anesthesiology and Critical Care Medicine, Paris, France) for their precious help.