Levy, Richard J.^*,†; Vijayasarathy, C.^‡; Raj, Nichelle R.^*; Avadhani, Narayan G.^‡; Deutschman, Clifford S.^*

INTRODUCTION

Sepsis, the systemic inflammatory response syndrome (SIRS) and the multiple organ dysfunction syndrome (MODS) are the most common causes of death in noncardiac intensive care units in the United States and Europe (¹). To date, the pathophysiologic alterations underlying the organ-system abnormalities in these disorders are unknown. Cardiovascular dysfunction, involving both systolic and diastolic abnormalities, is frequently noted in sepsis, SIRS, and MODS (²). Key among the characteristic abnormalities is ventricular dilatation and depressed contractility. The latter is manifest clinically as a decrease in ejection fraction with a requirement for high levels of preload to maintain normal stroke volume (³). Although a number of mediators and pathways have been shown to be associated with myocardial depression in sepsis, a unifying cause has yet to be found. Because a decrease in bioenergy available for contraction is a potential etiology, prior work has evaluated high-energy phosphate levels in a number of tissues. Some investigators have found preserved cardiac and skeletal muscle ATP levels during sepsis (^4–8). Other studies demonstrated decreases in myocardial ATP and skeletal myocyte phosphocreatine levels (^6,9,10). However, none of these studies has directly examined the integrity of the process of mitochondrial respiration. Because one hallmark of sepsis is an inability to consume or use molecular oxygen, it has been proposed that a defect in oxidative phosphorylation underlies sepsis-associated organ dysfunction (¹¹).

The process of oxidative phosphorylation involves the oxidation of NADH and FADH₂ generated by the Krebs cycle and the β-oxidation of fatty acids. This is catalyzed by a series of five enzyme complexes that comprise the respiratory chain on the mitochondrial inner membrane. In this process, electrons are transported from one enzyme to the next to ultimately generate ATP. Proton pumps within complexes I, III, and IV maintain a transmembrane hydrogen ion gradient that is integral to ATP generation. Dysfunction of any of the five complexes could impair oxidative phosphorylation and reduce the potential energy of the system. In the setting of sepsis, the enzyme complex most intriguing to study is complex IV, also called cytochrome c oxidase. Cytochrome c oxidase is the terminal oxidase in the electron transport system. It catalyzes the reduction of molecular oxygen to water using electrons from reduced cytochrome c generated by complex III. The consumption of oxygen in this process is closely linked to proton pumping across the inner mitochondrial membrane. Complex V then uses the trans-membrane proton gradient for ATP generation. Because of the documented defect in oxygen extraction in sepsis, examination of the kinetic activity of complex IV is logical. Therefore, in this study, we test the hypothesis that sepsis, induced by cecal ligation and puncture (CLP) in mice, alters the kinetics of cytochrome c oxidation by complex IV.

Cytochrome c oxidase contains 13 subunits. Subunit I, the active site, catalyzes the key activity of the complex via the heme a,a₃ binuclear center. Electrons from reduced cytochrome c are transferred from the divalent copper center in subunit II to the heme a center of subunit I and are then turned over to the heme a₃ center, where oxygen is reduced to water. Previous studies have demonstrated that sepsis alters transcription in a number of tissues (^12–14). Because the major activities of cytochrome c oxidase are performed by subunit I, we chose to test the secondary hypothesis that sepsis impairs the expression of subunit I of cytochrome c oxidase and the abundance of heme a,a₃.

MATERIALS AND METHODS

Induction of sepsis

All animal studies were approved by the University Laboratory Animal Resources committees at the University of Pennsylvania and conformed to National Institutes of Health standards. Under isoflurane general anesthesia, 6- to 8-week-old male C57B1/6 mice (Charles River Laboratories, Boston, MA) underwent sham operation, single-puncture CLP (CLP), or double-puncture CLP (2CLP) as previously described (^12–17). In the sham-operated mice, the cecum was manipulated but was not ligated or punctured. Punctures in CLP and 2CLP were performed with a 23-gauge needle. All animals were administered 50 mL/kg saline subcutaneously immediately postprocedure and then every 24 h. Once awake, animals were given access to food and water ad libitum. Animals were sacrificed at 0, 3, 6, 16, 24, or 48 h postprocedure. Three mice per time point per group were studied. Because of the mortality associated with 2CLP (50% at 24 h, 75% at 48 h, and 90% at 72 h), six mice were randomized to the 2CLP 24-h group and 12 mice were designated as 2CLP 48-h time points. Under deep pentobarbital anesthesia (50 mg/kg) and before apnea, the heart was excised and cardiac ventricles were harvested immediately. RNA was isolated in a separate group of animals.

Cytochrome c oxidase steady-state kinetics

Cardiac ventricles were homogenized in H medium (70 mM sucrose, 220 mM mannitol, 2.5 mM HEPES, pH 7.4, and 2 mM EDTA). Mitochondria were isolated by differential centrifugation and mitochondrial protein concentration was determined (¹⁸).

Cytochrome c oxidase kinetics were assayed by the method of Smith in which the rate of oxidation of ferrocytochrome c is measured by following the decrease in absorbance at 550 nm (^18,19). Assays were executed in a 1-mL reaction volume containing 50 mM PO₄⁻² (pH 7.0), 2% lauryl maltoside, and 0.5 μg of mitochondrial protein. Ferrocytochrome c was added at concentrations of 80, 40, 20, 10, 5, and 2 mM to initiate the reaction. First-order rate constants were calculated from mean values of three to four measurements at each ferrocytochrome c concentration. Specific activities were calculated using 21.1 mM⁻¹ cm⁻¹ as the extinction coefficient of ferrocytochrome c at 550 nm and Lineweaver-Burk plots were constructed. V_max values were determined from the y axis intercepts and K_m values were determined from the x axis intercepts.

Measurement of heme a,a₃ content

Mitochondria were isolated from cardiac ventricles via differential centrifugation. Mitochondrial heme a,a₃ content was calculated from the difference in spectra (dithionate/ascorbate reduced minus ferricyanide oxidized) of mitochondria solubilized in 10% lauryl maltoside using an absorption coefficient of 24 mM⁻¹ cm⁻¹ at 605 to 630 nm as described by Vijayasarathy et al. (¹⁸).

Northern blot hybridization of cytochrome c oxidase subunit I

Total RNA was extracted from cardiac tissue using the method of Chomczynski and Sacchi (²⁰). As previously described (^{12–14,17,21}), 10 μg of denatured RNA was resolved by electrophoresis on 1.5% formaldehyde-containing agarose gels. Gels were transferred to nylon membranes and were hybridized against purified, double-stranded ³²P-labeled cytochrome oxidase subunit I cDNA probes. Autoradiography and phosphorimaging were performed and data were normalized to the density of the 18S ribosomal subunit.

Protein immunoblotting

10-μg samples of mitochondrial protein were subjected to SDS-acrylamide gel electrophoresis and immunoblotting as previously described (¹³). Blots were labeled with a primary polyclonal antibody to mouse cytochrome c oxidase subunit I (Molecular Probes, Eugene, OR) and were secondarily exposed to rabbit anti-mouse IgG (Santa Cruz Biotechnology, Santa Cruz, CA). The signal was detected with enhanced chemiluminescence (ECL; Amersham Pharmacia Biotech, Piscataway, NJ), and density was measured using scanning densitometry.

Statistical analysis

Three animals per time point were evaluated. Data is presented as the mean ± standard deviation. Best-fit linear regression was applied to each Lineweaver-Burk plot. Statistical significance was assessed using the student's test with P < 0.05.

RESULTS

Cytochrome c oxidase inhibition

Data from steady-state kinetic assays were used to construct Lineweaver-Burk plots for different time points after sham operation, CLP, and 2CLP. These plots for 6, 16, 24, and 48 h postintervention are depicted in Figure 1. After sham (blue) and CLP (black), the slope of the best-fit line and K_m increased at 6, 16, and 24 h compared with baseline (time zero, nonoperative control, green line). V_max, however, remained unchanged throughout. For both, this represented a pattern of competitive (reversible) inhibition. By 48 h after sham operation, slope and K_m returned to baseline. Beginning 24 h after CLP, slope and K_m moved toward, but never reached, baseline values. Results after 2CLP were quite different (red lines). Up to 24 h after 2CLP, slope and K_m increased in a manner similar to that observed after sham operation and CLP, consistent with competitive inhibition. At 48 h, however, whereas the slope remained increased and the K_m approached baseline, V_max significantly decreased (Table 1). This represented a pattern of noncompetitive or irreversible inhibition.

Table 1

Fig. 1

Heme a,a₃ content decreases late in sepsis

One explanation for irreversible inhibition is the destruction of the heme catalytic center leading to decreased binding in a manner that cannot be reversed. Therefore, heme a,a₃ content in cytochrome c oxidase subunit I was examined. Heme a,a₃ content decreased significantly 48 h after 2CLP (Fig. 2), whereas values after sham operation and CLP were unchanged.

Fig. 2

Steady-state levels of cytochrome c oxidase subunit I decrease in sepsis

Previous studies in liver have demonstrated a persistent decrease in the expression of key proteins in liver and lung after 2CLP (^{12–14,16,17}). Persistently decreased expression of subunit I could also explain the alteration in V_max observed at 48 h after 2CLP. Therefore, Northern blot analysis was used to examine steady-state levels of mRNA encoding subunit I (Fig. 3). After sham operation, there was a transient decrease in subunit I mRNA levels 3 h after intervention. From 6 h onward, levels after sham operation were equivalent or higher than values at T₀. After CLP, mRNA levels decreased but returned to baseline by 16 h after operation. This is similar to observations in the liver (^{12–14,16,17}). In contrast, steady-state levels of subunit I mRNA decreased and remained depressed at all time points after 2CLP. Again, this is consistent with observations in lung and liver (^{12–14,16,17}).

Fig. 3

Protein expression parallels changes in mRNA

For altered steady-state mRNA levels to be meaningful, protein levels must be altered also. Therefore, immunoblotting was used to examine cytochrome c oxidase subunit I abundance (Fig. 4). After sham operation, levels decreased early and approached baseline by 24 h. The decrease in levels after CLP and 2CLP was persistent and more pronounced in double-puncture animals during the late phase (Fig. 4).

Fig. 4

DISCUSSION

The studies presented here reveal a fundamental, sepsis-associated biochemical defect in the mitochondrial pathway that leads to the reduction of molecular oxygen to water and the production of high-energy phosphates. We demonstrate explicitly, in myocardial tissue, an irreversible, noncompetitive block in the ability of cytochrome c oxidase, complex IV in the electron transport chain, to use cytochrome c late in fulminant sepsis. This correlates with a loss of the catalytic heme a,a₃ moiety and a decrease in the abundance of subunit I of cytochrome c oxidase, where this heme group resides. Furthermore, these findings correlate with decreased expression of the gene encoding subunit I. These results corroborate work done by Wei and coworkers (²³) that demonstrated nitric oxide-dependent decreases in cytochrome oxidase subunit I mRNA and protein in lipopolysaccharide-treated macrophages. Our findings have important implications with regard to sepsis-associated myocardial dysfunction and perhaps failure in the other organ systems affected by the disorder.

In previous studies, we and others have used the CLP model in rodents to investigate a number of sepsis-associated abnormalities in different organ systems (^15,17,24). After CLP, mice and rats develop inflammation, but recovery is the rule (^14,17,21). In contrast, 2CLP results in a predictable mortality of 50% at 24 h, 75% at 48 h, and 85% to 90% at 72 h postprocedure (^17,21). Chaudry and coworkers (²⁵) have characterized cardiodynamic dysfunction in this model, showing that, up to about 20 h after 2CLP, cardiac output is increased. At subsequent time points, however, contractility and cardiac output decline. These sepsis-associated alterations in cardiac performance are well documented, but the etiology of the hypodynamic state remains obscure. Our findings offer a potential explanation: an initial reversible and later irreversible inability to generate the energy required for contraction and cellular homeostasis. In addition, although the cause of death in this model of sepsis is unknown, several investigators have speculated that it is secondary to cardiovascular collapse (²⁶). The irreversible defect in myocardial ATP generation late in the course of severe sepsis correlates temporally with the development of late myocardial depression, supporting this proposal. Indeed, the irreversible impairment in cytochrome c oxidase function and, as a result, in ATP generation, also correlates with the high mortality seen after 2CLP and contrasts with the recovery characteristic of CLP.

One limitation of our work is the lack of a functional correlate to the kinetic abnormalities and alterations in expression of cytochrome oxidase we have reported. Chaudry's work (^25,27) compared 2CLP with SO in rats and mice at 5 and 20 h postprocedure. This work has been generally accepted and universally cited. Because our methodology duplicates that of Chaudry et al. almost exactly, we believe that the function should be the same. Nevertheless, it will be important to demonstrate a correlation between altered activity and expression of cytochrome oxidase and cardiac performance at each time point after 2CLP, CLP, and SO.

There are several potential explanations for a decrease in the V_max for cytochrome c oxidase 48 h after 2CLP. We specifically investigated two of them. Because we examined specific activity per milligram of total mitochondrial protein, a decrease in the quantity of enzyme as a component of mitochondrial protein could explain the alteration. This is borne out by data on steady-state levels of subunit I mRNA and protein. Alternatively, the change in V_max could result from the generation and binding of a noncompetitive or irreversible inhibitor. Such irreversible inhibitors of cytochrome c oxidase include carbon monoxide (CO), cyanide (CN⁻), lipid peroxidation due to reactive oxygen species, and peroxynitrite. Increases in CO, reactive oxygen species, and peroxynitrite have been reported in sepsis (^28–30). A recent study has identified a mechanism for endogenous CN⁻ generation in the brain, but this has not been extended to myocardium (³¹). In sepsis, heme oxygenase-1 (HO-1) is induced and leads to the generation of CO by catalyzing the breakdown of heme to CO and biliverdin (^32–34). The active site of cytochrome c oxidase, containing a heme a,a₃, could certainly be a target of HO-1 in sepsis. Similarly, reactive oxygen species directly peroxidize lipids and have the potential to destroy the functional integrity of cytochrome c oxidase. The net effect of these destructive processes would be a decrease in the quantity of heme a,a₃ and/or cytochrome c oxidase as a result of degradation. The data in Figures 2 and 4 support this. Alternatively, our previous studies on lung and liver after 2CLP have revealed a decrease in the expression of key proteins (^{12–14,16,17}). Failure to express subunit I of cytochrome c oxidase, as demonstrated here (Fig. 3), would also lead to decreased enzyme levels and activity. This also is consistent with previous findings in liver and lung. Thus, our findings support a loss of the active subunit of cytochrome c oxidase and, in particular, of the heme a,a₃ center, as a potential cause of failed enzyme activity. It should be pointed out that the gene encoding Complex IV subunit I is found in mitochondrial DNA. Therefore, the data presented here represent a unique example of an alteration in expression of mitochondrial DNA in the setting of an acute disease.

Prior studies of cardiac dysfunction in sepsis have measured ATP levels in dysfunctional myocytes. The results are difficult to interpret. However, the decreased ability to aerobically generate ATP, as would result from destruction of cytochrome c oxidase, need not result in a loss of ATP. If an alternative mechanism for ATP production exists, contractility and homeostasis, although depressed, could be maintained. Furthermore, if cardiomyocyte metabolism becomes down-regulated, contractility would decrease due to decreased energy consumption with preservation of ATP levels. Indeed, the decrease in contractility observed in sepsis might represent a mechanism to preserve high-energy phosphate levels. A process whereby cardiomyocyte metabolism becomes down-regulated with decreased contractility and preserved ATP levels has been observed in hypoxia and ischemia (²²). This process has been termed “myocardial hibernation.” In fact, it has been hypothesized that cytochrome c oxidase is the “sensor” that initiates hibernation in the face of decreased oxygen delivery to the mitochondria (²²). This process could well be occurring in sepsis.

Similar dysfunction in tissues other than the heart could easily reflect a loss of integrity of mitochondrial function. For example, we have recently demonstrated regenerative failure in the liver of septic mice (³⁵). Regeneration would clearly be compromised by a defect in oxidative phosphorylation. Further studies will examine specific etiologies of myocardial cytochrome c oxidase impairment, examine other enzymes in the pathway, and seek to demonstrate a defect in other organ systems. A unified mechanism to explain the diverse abnormalities observed in sepsis, which has thus far eluded identification, may emerge.

REFERENCES