Discussion
This study is the first to demonstrate mitochondrial dysfunction in skeletal muscle of patients with protracted critical illness. In the skeletal muscles of these patients we observed approximately 50 % reduction in the ability to synthetize ATP by aerobic phosphorylation per mg of muscle wet weight (OXPHOS/W
w) which correlated with the concentration of depleted complex IV. Complex III was also depleted, unlike complexes II and V. When OXPHOS was adjusted to citrate synthase activity (OXPHOS/CS), the differences between ICU patients and control subjects disappeared and OXPHOS/W
w strongly correlated with citrate synthase activity. The obvious interpretation of these results is that mitochondria are depleted in ICU patients, whilst complexes II and V are relatively abundant in remaining functional mitochondria. A similar disproportionality of the concentrations of respiratory complexes has been described in skeletal muscle during aging [
16] and oxidative stress [
17]. Even though citrate synthase activity is widely used as a marker of mitochondrial content [
2,
18‐
20], it may become a subject of oxidative damage [
21] and therefore it may not reliably reflect the mitochondrial density. Because we have not used an alternative method of measuring mitochondrial content (e.g., electron microscopy), we cannot say whether the depletion of complexes III and IV occurred in isolation or as part of mitochondrial depletion. It is the concentration of the depleted complex IV (and possibly complex I) that was limiting for the mitochondrial function, in keeping with data of Levy [
22], who demonstrated the relation of complex IV dysfunction to bioenergetics failure in acute sepsis. Contrary to our hypothesis, there was no sign of increased mitochondrial uncoupling in ICU patients.
In order to explore the functional capacity of individual complexes, we performed a respirometry protocol in which we used specific substrates and inhibitors of individual complexes. If expressed per muscle wet weight (Table
2), we saw a trend towards increase in functional capacity of respiratory complexes II and III, whilst that of complexes I and IV tended to be non-significantly reduced to approximately 70 % of values seen in control subjects, and correlated with OXPHOS. After adjustment for citrate synthase activity, complexes II and III were increased significantly (threefold and twofold respectively,
p <0.01) and complexes I and IV were not different (Fig.
3). High-resolution respirometry measures the changes in oxygen consumption in fresh intact tissue homogenates after addition of respiratory substrates and inhibitors [
11]. The sample contains intact mitochondria in a cytosolic context and it is believed that this approach better reflects physiological alterations occurring in vivo [
23]. The technique has been calibrated against permeabilized muscle fibers [
12] and isolated mitochondria [
8]. When using this method for measuring the functional capacity of individual complexes one must bear in mind that the rate-limiting step can in theory appear downstream of the complex that is being analyzed. Complexes III and IV are under physiological conditions able to accommodate the flux of electrons from both complexes I and II and it is therefore unlikely that they become rate-limiting when fed by electrons from either complex I or II in isolation. For testing complex III we used glycerol-3-phosphate as a substrate whilst complexes I and II had been blocked. By doing so we avoided the risk of downstream limitation (i.e., at complex IV), but on the other hand, the rate-limiting step may be at the level of GPDH, which is functionally a part of the glycerol phosphate shuttle rather than the respiratory chain.
With these limitations of respirometry in mind, we repeated the measurements of individual complex activities by a different technique. Classical spectrophotometry is a well-established method [
2,
3,
18], which assesses the activities of respiratory complexes by using artificial complex-specific substrates after the organelle structure has been destroyed by repeated freezing and thawing. This means that the measured activity of each complex is independent of the functionality of other complexes. As demonstrated in Fig.
3, both methods gave very similar results and confirmed the increased functional capacity of complexes II and III/GPDH in the critically ill as compared to control subjects.
Complex II (succinate dehydrogenase) normally drives electrons from succinate oxidation to fumarate in the citric acid cycle (CAC) via flavin adenine dinucleotide (FAD) to the respiratory chain. CAC itself is heavily dependent on reoxidation of NADH by complex I as it produces three molecules of NADH per one molecule of FADH
2. Eventual increase in NADH/NAD+ ratio inhibits CAC. Similarly, aerobic glycolysis produces 2NADH/molecule of glucose during the conversion to pyruvate and a further 2NADH by converting pyruvate to acetyl-CoA, which is oxidised in CAC. However, during oxidation of fatty acid and carbon skeletons of branched chain amino acids, reduced coenzymes FADH
2 and NADH are produced in a 1:1 ratio. Of all catabolic pathways, fatty acid oxidation is thus least dependent on the functionality of complex I. In the acute phase of critical illness complex I seems to be predominantly impaired [
2] and upregulation of complex II at a later stage can be a compensatory response or an attempt to bypass dysfunctional complex I. Insulin resistance is a well-known feature of critical illness [
24,
25] and it has been shown that GLUT-4 dependent transport is dysfunctional in patients with ICUAW (weakness developing in a critically ill patient without an identifiable cause other than nonspecific inflammation) [
26] and pyruvate dehydrogenase is inhibited [
27]. Skeletal muscle in protracted critical illness thus may suffer from starvation of carbohydrate-derived substrate for CAC. On the contrary, free fatty acids are elevated in the critically ill [
24,
28] and intracellular lipid droplets accumulate early in diaphragmatic and biceps muscle in brain-dead donors [
18]. Branched-chain amino acids (BCAA) derived from muscle protein degradation are deaminated in skeletal muscle and their carbons are oxidized in a similar way to fatty acid oxidation. Relative upregulation of complex II in the context of mitochondrial dysfunction may thus represent an adaptive response to insulin resistance [
29] and preferential oxidation of lipids and BCAA over carbohydrates. Glycerol-3-phosphate can be formed from glycerol derived from lipolysis [
30], and it requires respiratory complexes distal to complex I to be converted to glyceraldehyde-3-phosphate [
31], a glycolytic intermediate. Upregulation of complex III/GPDH seen in our ICU patients may reflect the increase in intracellular lipid turnover in the skeletal muscle of these patients.
However, the lack of correlation between OXPHOS and both functional capacities and relative abundance of complexes II and III/GPDH suggests that they may play other functions, which are not directly related to aerobic ATP production. It has been recently shown that cells accumulate succinate during hypoxia [
32‐
34] or inflammation [
35]. When oxygenation is restored, rapid re-oxidation of succinate produces electron flux, which downstream complexes are unable to absorb, and which is redirected backwards to complex I, generating excessive amounts of reactive oxygen species [
36,
37]. Relative redundancy of the activity of complexes II and III over complex I observed by us in protracted illness could be an adaptation against cell damage when intracellular succinate levels are fluctuating.
Indeed our study has many limitations. First, our data are derived from a small group of highly selected subjects. We found it very difficult to consent patients for the biopsy in this non-therapeutic study. With such a small number of subjects there is always a risk of type II error, i.e., that we were unable to detect changes that were present. High inter-individual variability in the concentration and functionality of respiratory complexes (see Fig.
3) is well-known [
2,
22], and further complicates the interpretation of data. Biopsies were performed in ICU patients who had been ventilator-dependent for more than 2 weeks (mean 28 days) and suffered from muscle weakness. We have selected this cohort of patients with muscle dysfunction in order to maximize the chances of observing any alteration of bioenergetics in a non-respiratory muscle, which seems to be less affected, even in the acute phase of critical illness, when compared to the diaphragm [
18,
38] or intercostal muscles [
3]. As a result, it remains unclear whether the changes in mitochondrial metabolism described above are consequences of prolonged immobility [
39‐
41], the critical illness, or whether they occur only in patients who are weak. Of note, our control subjects were ambulatory elective hip surgery patients and it is unknown whether their potentially reduced mobility affected the mitochondrial function of skeletal muscle. In light of this, our pilot study should be treated as a proof-of-concept study and the results interpreted with caution.
Competing interests
The authors declare no competing interests.
Authors’ contributions
KJ processed muscle samples, participated in the respirometry analysis and contributed to the study design. AK and JZ helped to design the study, processed muscle samples, performed (together with MK) the respirometry analyses. ME and JT (together with AK) performed spectrophotometric analysis of respiratory complexes. VD, MF and JG obtained informed consents and performed the biopsies, whilst VFN and JK carried out the western blots. FD conceived of the study, participated in its coordination and performed the statistical analysis. All authors wrote their parts of the manuscript, revised the first draft and then read and approved the final version of the manuscript.