Patients with end-stage heart failure (HF) have respiratory complications that contribute to disease morbidity and mortality. The incidence of pneumonia in end-stage HF patients who receive a heart transplant is 15–20% and 5% of patients have prolonged respiratory failure post-heart transplant (Lenner et al., 2001). The cause of pneumonia and respiratory failure in end-stage HF patients is multifactorial. In this context, inspiratory muscle abnormalities in patients with end-stage HF can play a critical role in respiratory complications (Kelley and Ferreira, 2016). For instance, pre-operative inspiratory muscle strength training reduces the incidence of post-operative pulmonary complications in “high-risk” patients, including those with HF, undergoing coronary artery bypass graft (Hulzebos et al., 2006).

The primary inspiratory muscle is the diaphragm, which is necessary for normal ventilatory and expulsive behaviors that promote airway clearance (Sieck and Fournier, 1989; Mantilla and Sieck, 2013). Diaphragm dysfunction has been documented in animal models of HF, being characterized by atrophy and contractile impairments that diminish force and power capabilities (Howell et al., 1995; Stassijns et al., 1998; van Hees et al., 2007; Ahn et al., 2015; Kelley and Ferreira, 2016). The diaphragm of end-stage HF patients also shows ultra-structural and myofibrillar protein alterations that suggest metabolic and contractile abnormalities (Lindsay et al., 1996). However, less is known about the underlying mechanisms of diaphragm dysfunction in end-stage HF.

Excess reactive oxygen species (ROS) and redox imbalance play a causal role in diaphragm dysfunction in animal models of HF (Supinski and Callahan, 2005; Ahn et al., 2015). An important source of ROS in the diaphragm is the Nox2 isoform of NAD(P)H oxidase (Pal et al., 2013; Loehr et al., 2014; Bost et al., 2015). A functionally assembled Nox2 enzyme complex consists of several subunits (Nox2, p22^phox, p67^phox, p40^phox, Rac1/2, and p47^phox), and enzyme activation requires p47^phox phosphorylation (Javesghani et al., 2002; Lassègue et al., 2012; Pal et al., 2013; Ferreira and Laitano, 2016). Diaphragm p47^phox phosphorylation is increased in mice with HF, and genetic deletion of p47^phox prevents excess ROS release and contractile dysfunction in diaphragm of mice with HF (Ahn et al., 2015). These findings in mice suggest a pivotal role for Nox2-derived ROS on diaphragm dysfunction in HF.

Diaphragm antioxidant enzymes are unchanged (Ahn et al., 2015) or elevated in animal models of HF (Bowen et al., 2015b; Mangner et al., 2015). The increase in antioxidant enzymes may reflect a compensatory response to scavenge excess ROS. However, these compensatory responses appear insufficient to maintain cellular redox homeostasis as markers of protein oxidation are elevated in diaphragm of HF animals (acute Bowen et al., 2015a and chronic models Supinski and Callahan, 2005; Coirault et al., 2007). This is relevant because heightened ROS and protein oxidation promote diaphragm atrophy and impair contractile function (Ferreira and Reid, 2008; Powers et al., 2011).

Despite advances in understanding the causes and role of redox imbalance on diaphragm dysfunction in animal models of HF, very little is known about ROS sources, antioxidant enzymes, and protein oxidation in the diaphragm of patients. We tested the hypotheses that diaphragm from patients with end-stage HF show heightened Nox2 subunit levels and p47^phox phosphorylation with unchanged protein levels of antioxidant enzymes which results in elevated markers of protein oxidation.

Patient characteristics are detailed in Table 1. In summary, patients exhibited HF caused by ischemic (n = 5) and non-ischemic cardiomyopathy (n = 6).

Diaphragm mRNA levels of Nox2, p22^phox, p47^phox, p67^phox, and p40^phox were increased with median values ranging from 2.5- to 4.6-fold over controls (Figure 1). The protein levels of Nox2, p47^phox, p22^phox, and p67^phox were also increased in diaphragm of HF patients, while protein levels of Rac1 was not significantly changed (Figure 1). We were not able to detect p40^phox via immunoblot in the diaphragm, which is consistent with a previous study (Javesghani et al., 2002).

Phosphorylation of p47^phox promotes activation of Nox2 (El-Benna et al., 2009; Lassègue et al., 2012), thus we examined the phosphorylation status of p47^phox using antibodies against specific phosphorylated serine residues. We found that phosphorylation at Ser328, Ser345, Ser359, and Ser370 were increased in diaphragm of HF patients compared to controls when normalized to total protein (Figure 2). When we normalized the phosphorylated signal from each serine residue to the total p47^phox signal, there was no difference in phospho-to-total p47^phox between control and HF patients (Figure 2). This suggests that the total abundance of p47^phox protein in the phosphorylated state was elevated, whereas the “percentage” of phosphorylated p47^phox was unchanged.

We measured the protein level of key cytosolic and mitochondrial antioxidant enzymes (Figure 3). Catalase levels were increased 2.2-fold over controls (P < 0.05), whereas the levels of glutathione peroxidase (P = 0.15), SOD1 (P = 0.48), and SOD2 (P = 0.22) were unchanged in the diaphragm of HF patients.

Despite increased levels of catalase, redox imbalance in diaphragm from patients was manifested by increased (~40%) protein carbonyls, while 4-HNE and 3-NT were unchanged in the diaphragm of end- stage HF patients (Figure 4).

Our study shows that diaphragm of patients with end-stage HF have elevated mRNA and protein levels of Nox2 subunits that is accompanied by increased p47^phox phosphorylation, which is consistent with Nox2 activation. The antioxidant enzyme catalase was also increased in diaphragm of patients, while superoxide dismutases and glutathione peroxidase were unchanged. These findings suggest disrupted redox homeostasis in the diaphragm of patients with end-stage HF, which are confirmed by elevated levels of protein carbonyls.

The enzyme Nox2 is emerging as an important source of oxidants that cause diaphragm abnormalities in animal models of diseases, including muscular dystrophy (Whitehead et al., 2010; Pal et al., 2014; Henriquez-Olguin et al., 2015) and HF (Ahn et al., 2015). The functionally assembled Nox2 complex includes several subunits (Nox2, p47^phox, p22^phox, p40^phox, p67^phox, and Rac1 Bedard and Krause, 2007; Lassègue et al., 2012). We observed that mRNA levels of several Nox2 subunits were increased in the diaphragm of end-stage HF patients, with p47^phox subunit having the highest elevation (Figure 1).

Increased p47^phox mRNA was translated into higher protein abundance compared to control (Figure 1). Similarly, we have found heightened protein levels of p47^phox in the diaphragm of mice with HF (Ahn et al., 2015). It is unclear whether elevated p47^phox is sufficient to heighten Nox2 activity in skeletal muscle cells. Overexpression of p47^phox increases Nox2 activity in glial cells (Lavigne et al., 2001). Thus, it is possible that skeletal muscle cells have a constitutive, p47^phox-dependent Nox2 activity.

The canonical pathway for Nox2 activation involves p47^phox phosphorylation at serine residues that releases auto-inhibition of membrane- and subunit-binding domains (El-Benna et al., 2009; Drummond et al., 2011; Lassègue et al., 2012). The C-terminal domain of p47^phox contains 11 serine residues between amino acids 303–379 that encompasses the auto-inhibitory region. Point mutations have revealed six serine residues of p47^phox that are required for full activation of Nox2: Ser303, Ser304, Ser328, Ser359, Ser370, and Ser379 (reviewed in El-Benna et al., 2009). We found that total levels of phosphorylated Ser328, Ser345, Ser359, and Ser370 were elevated in diaphragm of patients with end- stage HF (Figure 2). When we calculated the phospho-to-total p47^phox ratio, differences between HF and controls were not statistically significant (Figure 2C). These findings suggest that increased “absolute” levels of phosphorylated p47^phox accompanied the heightened expression of total p47^phox. However, the “percentage” (or relative levels) of p47^phox protein in the phosphorylated state was unchanged. Typically, an increase in relative levels/percentage of p47^phox phosphorylation (i.e., elevated phospho-to-total p47^phox) is considered an indicator of Nox2 activation (Isabelle et al., 2005). In the context of elevated levels of total p47^phox, unchanged phospho-to-total p47^phox ratio should also heighten Nox2 activation because phosphorylated p47^phox proteins are more abundant. Biologically, the absolute amount of phosphorylated p47^phox would dictate Nox2 activity. Therefore, we consider that Nox2 activity is likely increased in diaphragm of end-stage HF patients. However, the unchanged relative levels/percentage of phospho-p47^phox has implications regarding mechanisms of Nox2 activation. Our data suggest that the activity of kinases that phosphorylate p47^phox is not necessarily elevated in diaphragm of patients with end-stage HF. It is possible that other pathways of p47^phox signaling (e.g., arachidonic acid) or overexpression per se mediate Nox2 activation in the diaphragm (Ferreira and Laitano, 2016). Alternatively, enhanced diaphragm p47^phox phosphorylation may be involved in the pathophysiology of diaphragm dysfunction at earlier stages of the disease.

We have not tested Nox2 activity in our study because our tissue collection method (flash freezing) does not lend the sample suitable for reliable measurements of activity, as per recent recommendations (Rezende et al., 2016). We have found, in diaphragm of mice with HF, increases in total p47^phox and phospho-to-total p47^phox that are consistent with increases in Nox2 activity (Ahn et al., 2015). Indeed, knockout of p47^phox prevented excess diaphragm ROS emission suggesting Nox2 as a major source of pathological diaphragm oxidants in HF (Ahn et al., 2015). Overall, our data in humans and animals suggest elevated Nox2 activity in patients with HF.

A decrease in protein levels or intrinsic activity of antioxidant enzymes will contribute to ROS accumulation that disrupts cellular redox balance. Major intracellular antioxidant enzymes include catalase, glutathione peroxidase, and superoxide dismutases (SOD1 and SOD2). Patients had increased levels of catalase, whereas there was no statistical difference in the levels of GPX, SOD1, and SOD2 between patients and controls (Figure 3). This outcome is likely due to our limited sample size and large variability in the human diaphragm data. In animal models of HF with reduced ejection fraction, diaphragm levels of SOD1 or SOD2 were either unchanged (Ahn et al., 2015; Laitano et al., 2016) or elevated (Mangner et al., 2015), whereas the activity of GPX was increased (Mangner et al., 2015) and catalase was unchanged (Mangner et al., 2015). Nonetheless, diaphragm catalase activity was increased in a model of HF with preserved ejection fraction (Bowen et al., 2015b). In general, our data in patients and studies in animals suggest that heightened protein oxidation in the diaphragm induced by HF cannot be explained by a decrease in the protein levels or activity of antioxidant enzymes. Heightened diaphragm antioxidant enzyme levels in HF might be a compensatory adaptation aimed to maintain redox balance when ROS production is increased. However, our findings suggest that any compensatory response is insufficient to maintain normal protein oxidation levels in patients with end-stage HF.

Protein carbonyls, a marker of oxidation, were elevated in the diaphragm of end-stage HF patients (Figure 4). These results corroborate previous findings in animals with severe HF induced by aortic stenosis (Coirault et al., 2007) or in the early stages post-myocardial infarction (Supinski and Callahan, 2005; Bowen et al., 2015a), but disagrees with data from our group in rats and mice with moderate HF in the later stages post-myocardial infarction (Ahn et al., 2015; Laitano et al., 2016). We speculate that the increase in diaphragm protein carbonyls in chronic HF is related to disease severity. This concept is supported by progressive increases in systemic markers of oxidation in groups of patients going from NYHA Class I to IV (Belch et al., 1991; Nishiyama et al., 1998).

Excess oxidation can contribute to diaphragm abnormalities due to protein degradation as well as impaired function of excitation-contraction coupling and sarcomeric proteins. Oxidation enhances protein degradation by calpain, caspase-3, and the proteasome (Grune et al., 2003; Moylan and Reid, 2007; Smuder et al., 2010). Calpain and proteasome activity are elevated in diaphragm of HF rats (Dominguez and Howell, 2003; van Hees et al., 2008a), and proteasome inhibition prevents myofibrillar protein degradation and attenuates diaphragm weakness in HF rats (van Hees et al., 2008a). Regarding protein function, carbonylation of sarcomeric proteins in vitro impairs actomyosin cross- bridge kinetics and, in general, exposure to oxidants depresses diaphragm force and shortening velocity (Perkins et al., 1997; Callahan et al., 2001; Coirault et al., 2007). These functional outcomes are consistent with impaired diaphragm contractile properties in HF animals (Lecarpentier et al., 1998; Coirault et al., 2007; van Hees et al., 2007, 2008b; Empinado et al., 2014; Ahn et al., 2015), which are prevented by pharmacologic antioxidants (Supinski and Callahan, 2005; Laitano et al., 2016) or knockout of p47^phox (Ahn et al., 2015).

Diaphragm of patients with end-stage HF shows upregulation of Nox2 subunits, increased total but unchanged relative levels of phosphorylated p47^phox, and elevated abundance of catalase. These changes in ROS-producing and scavenging enzymes culminated in elevated diaphragm protein oxidation. Overall, our findings suggest that Nox2 is an important source of ROS in the diaphragm of patients with end-stage HF and increases in catalase levels are not sufficient to maintain cellular redox homeostasis.

Sample collection and processing: JM, CH, MB, and LF. Experiments: BA, PC, AB, JM, NP, and LF. Data analysis and interpretation: BA, PC, AB, JM, NP, MB, CH, AJ, and LF. Manuscript writing and editing: BA, PC, AJ, and LF. Patients assessment and surgeries: CH and MB.

This study was funded by NIH grant R00-HL098453 to LF. AJ was funded by NIH grant R01 AR060209.