|Home | About | Journals | Submit | Contact Us | Français|
Rationale: Hyperoxia-induced acute lung injury has been used for many years as a model of oxidative stress mimicking clinical acute lung injury and the acute respiratory distress syndrome. Excess quantities of reactive oxygen species (ROS) are responsible for oxidative stress–induced lung injury. ROS are produced by mitochondrial chain transport, but also by NADPH oxidase (NOX) family members. Although NOX1 and NOX2 are expressed in the lungs, their precise function has not been determined until now.
Objectives: To determine whether NOX1 and NOX2 contribute in vivo to hyperoxia-induced acute lung injury.
Methods: Wild-type and NOX1- and NOX2-deficient mice, as well as primary lung epithelial and endothelial cells, were exposed to room air or 100% O2 for 72 hours.
Measurements and Main Results: Lung injury was significantly prevented in NOX1-deficient mice, but not in NOX2-deficient mice. Hyperoxia-dependent ROS production was strongly reduced in lung sections, in isolated epithelial type II cells, and lung endothelial cells from NOX1-deficient mice. Concomitantly, lung cell death in situ and in primary cells was markedly decreased in NOX1-deficient mice. In wild-type mice, hyperoxia led to phosphorylation of c-Jun N-terminal kinase (JNK) and extracellular signal–regulated kinase (ERK), two mitogen-activated protein kinases involved in cell death signaling, and to caspase-3 activation. In NOX1-deficient mice, JNK phosphorylation was blunted, and ERK phosphorylation and caspase-3 activation were decreased.
Conclusions: NOX1 is an important contributor to ROS production and cell death of the alveolocapillary barrier during hyperoxia and is an upstream actor in oxidative stress–induced acute lung injury involving JNK and ERK pathways in mice.
Oxidative stress is a major feature of acute lung injury and the acute respiratory distress syndrome (ARDS) and may be responsible for pulmonary cell damage.
Targeting the NADPH oxidase NOX1 during hyperoxia-induced lung injury in mice inhibits oxidative-induced cell death pathway and lung damage, providing a new therapeutic approach in the treatment of acute lung injury.
Respiratory distress syndrome (RDS) of premature babies is characterized by immature lung development and insufficient surfactant production by alveolar type II cells. During RDS treatment with mechanical ventilation and high oxygen concentration, the lungs are exposed to increased oxidative stress leading to pulmonary injury (1). The damage to both endothelial and epithelial cells leads to the loss of alveolocapillary barrier integrity, resulting in interstitial edema and gas exchange impairment. We and others have previously shown that reactive oxygen species (ROS) are able to induce a cell death response with features of both apoptosis and necrosis during oxygen exposure (2, 3). The mechanisms of cell death are not completely defined in hyperoxia-mediated lung epithelial and endothelial cell death, but involve the activation of mitogen-activated protein kinase (MAPK) (4) and the caspase cascade (5). Inhibition of the c-Jun N-terminal kinase (JNK) or extracellular signal–regulated kinase (ERK) pathway, using dominant negative constructs or chemical inhibitors, prevents cell death in an epithelial cell line (MLE-12) and in human pulmonary artery endothelial cells during hyperoxia (6–9).
During supplemental oxygen therapy, ROS can be generated by a variety of enzymes such as mitochondrial chain transport and NADPH oxidase enzymes (10, 11). The well-known phagocyte NADPH oxidase complex includes a catalytic subunit, NOX2/gp91phox, which is associated with regulatory subunits (11). This complex is required for NOX activation and generation of superoxide (11, 12). The NOX2 complex plays an essential role in nonspecific host defense against pathogens (13). Several additional homologs of NOX2 have been identified: NOX1–NOX5, dual oxidase-1 (DUOX1), and DUOX2. Despite their similar structure and enzymatic function, NOX enzymes differ in their mechanism of activation. The NOX isoforms are expressed in several tissues and cell types, where they serve various physiological functions, including posttranslational modifications of proteins and cellular signal transduction (12, 14, 15). NOX enzymes are also involved in a variety of pathologies, including amyotrophic lateral sclerosis, hypertension, and lung emphysema (16–19).
A study demonstrated that both NOX1 and NOX2 are expressed in mouse lung, and NOX1 has been detected in a human alveolar epithelial cell line (A549) and in endothelial cells (20, 21). However, until now, no precise function of NOX1 and NOX2 in the lungs has been determined. In vitro studies demonstrated that diphenylenethidium (DPI), a wide inhibitor of NADPH oxidases, was effective in reducing hyperoxia-induced ROS generation and cell death in MLE-12 cells and in pulmonary endothelial cells (6, 9, 22). Using NOX1- and NOX2-deficient mice, we determined their specific contribution in hyperoxia-induced acute lung injury.
This study demonstrates for the first time that NOX1, but not NOX2, is an important mediator of acute lung injury mediated by oxidative stress, particularly in alveolar cells. NOX1-generated ROS are largely responsible for alveolar cell death and subsequent lung injury through JNK and ERK pathways during hyperoxia in mice. Some of these data have been reported in the form of an abstract (23).
See the online supplement for additional details on reagents and methods.
NOX1-deficient mice, NOX2 (gp91phox)-deficient mice, and wild-type (WT) mice were inbred on the C57BL/6J background (17, 24). Animals were kept under specific pathogen–free conditions. The animal procedure was performed in accordance with the Institutional Ethics Committee on Animal Care (Geneva, Switzerland) and the Cantonal Veterinary Office.
Eight- to 10-week-old female mice (deficient and WT) were exposed to room air or 100% O2 for 72 hours. Bronchoalveolar lavage (BAL) was performed as described (25). The total cell count was determined with a cell counter and protein concentration was measured. BAL fluid cell distribution was quantified in Cytospin preparations (Perbio Science, Lausanne, Switzerland) after staining with Diff-Quik dye (Dade Behring, Paris, France).
Paraffin-embedded sections of lung fixed in 4% paraformaldehyde were stained with hematoxylin–eosin or with anti-fibrin polyclonal antibody followed by EnVision G|2 System/AP (Dako, Glostrup, Denmark), or with an anti–cleaved caspase-3 polyclonal antibody followed by Texas red–conjugated secondary antibody. Double immunostaining was performed with anti-pJNK (phosphorylated JNK) or anti-pERK (phosphorylated ERK) monoclonal antibody combined with anti– SP-C (Chemicon, Zug, Switzerland) or anti–von Willebrand factor (VWF) followed by fluorescein isothiocyanate–conjugated secondary antibody before counterstaining with 4′,6-diamidino-2-phenylindole (DAPI). Slides were mounted with Fluor Save (Calbiochem, Darmstadt, Germany) and analyzed by confocal microscopy.
Alveolar macrophages were isolated from the BAL fluid of adult mice, and alveolar epithelial type II cells (type II AECs) and endothelial cells (MLECs) were isolated from adult mouse (8- to 14-wk-old) lungs as described with minor modifications (26–28). Hyperoxic conditions (24–72 h) were achieved by placing type II AECs and MLECs in a sealed glass chamber filled with 95% O2−5% CO2 at 37°C.
RNA was extracted with a total RNA isolation NucleoSpin RNA II kit (Macherey-Nagel, Düren, Germany) and reverse transcribed with SuperScript reverse transcriptase (SuperScript choice system; Invitrogen, Carlsbad, CA). We used the primers for NOX1, NOX2, and NOX4 and the reference genes encoding EEF1A1 (eukaryotic elongation factor-1A1) and HPRT (hypoxanthine–guanine phosphoribosyltransferase). For polymerase chain reaction primer sequences and details, see the online supplement.
Frozen lung tissues were cryosectioned (20 μm) and collected onto SuperFrost Plus slides (Perbio Science). To detect the various cell types that generated ROS, lung sections from all mouse strains were double immunostained with dihydroethidium (DHE) and anti–SP-C or with DHE and anti-VWF. Images were captured with an inverted microscope and analyzed with Metafluor imaging software (Molecular Devices, Basel, Switzerland). Quantification was performed by measuring the fluorescence intensity of all cells counted in the lung sections and of isolated type II and endothelial cells (>50 for each mouse) from three different mice.
Total lung protein extracts were performed as previously described (29). Total protein extracts were loaded and nitrocellulose membranes were blocked overnight in Tris-buffered saline–Tween buffer, 5% milk, followed by hybridization with anti-pJNK, anti-JNK, anti-pERK, anti-ERK, anti-actin, anti–cleaved caspase-3, or anti-poly(ADP-ribose) polymerase (PARP)-1 in the same buffer. The proteins were detected with enhanced chemiluminescence (ECL) reagents (Interchim, Montluçon, France). Densitometric evaluation was performed with Quantity One software (Bio-Rad, Reinach, Switzerland).
Lactate dehydrogenase was measured in BAL supernatant as described (30). For sections of fixed lung, terminal deoxynucleotidyltransferase–mediated dUTP nick end labeling (TUNEL) was performed with an in situ apoptosis detection kit according to the protocol of the manufacturer (Chemicon, Temecula, CA); and for primary cells, TUNEL was performed as described by the manufacturer (Roche, Indianapolis, IN). Slides were mounted with polyvinyl alcohol (Mowiol) and analyzed by confocal microscopy (LSM 510 Meta; Zeiss, Oberkochen, Germany). Quantification of positive staining was performed with Metamorph analysis software (Molecular Devices) (20 images per mouse, 3 mice per group, and 3 independent experiments).
For all parameters measured, the values for all samples under different experimental conditions were averaged and the SEM of the mean was calculated. The significance of differences between the values of the groups was determined by two-way analysis of variance with multiple comparisons followed by unpaired t test. The Wilcoxon rank sum test for unpaired data was also used. Significance levels were set at P < 0.05.
To address the role of NOX1 and NOX2 in hyperoxia-induced lung injury, we exposed NOX1- and NOX2-deficient mice, as well as WT mice, to 100% oxygen. In accordance with previous studies from our laboratory and others, WT mice exhibited features of severe lung damage after 72 hours of oxygen exposure. The extent of lung injury was not significantly different in NOX2-deficient mice compared with WT mice as measured by BAL protein content reflecting vascular leakage (Figure 1A); BAL lactate dehydrogenase release, an indicator of tissue injury (Figure 1B); and fibrin deposition in the alveolar spaces (Figures 1E–1G). In contrast, the absence of NOX1 significantly reduced hyperoxia-induced lung injury compared with WT mice (Figures 1C–1E and 1G). Most of the in vivo experiments with NOX1- and NOX2-deficient mice were performed independently, each of them with their respective WT control. Under normal conditions, lungs of NOX1- and NOX2-deficient mice displayed no morphological abnormalities and were similar to those of WT mice (data not shown).
We then evaluated hyperoxia-induced lung inflammatory responses in NOX1- and NOX2-deficient mice. Hyperoxia caused a significant increase in BAL total cell count (macrophages, lymphocytes, and neutrophils) that was similar in the three mouse strains (Table 1). BAL neutrophil counts of NOX1- and NOX2-deficient mice were higher compared with those from WT mice in hyperoxia, but the neutrophil count was similar in air-breathing animals (Table 1). These results indicate that NOX1, but not NOX2, participates in oxidative stress–induced acute lung injury independently of the acute lung inflammatory response.
To determine whether NOX1 and NOX2 are regulated by hyperoxia in type II AECs and in MLECs, we measured NOX1 and NOX2 expression. At baseline, NOX1 mRNA was expressed in type II AECs and MLECs (Figure 2A), in contrast to NOX2 mRNA, which was undetectable compared with the positive control. As positive control, we used alveolar macrophages in which NOX2 is strongly expressed, as expected (Figure 2B). After hyperoxia, NOX1 mRNA expression was increased in both type II AECs and MLECs whereas NOX2 was still undetectable in these primary cells (Figures 2A and 2B).
To determine the contribution of NOX1 in hyperoxia-induced superoxide formation, we measured ROS production in the lungs of WT and NOX1-deficient mice. Mice were exposed to air or hyperoxia for 72 hours and frozen lungs were cryosectioned. Lung sections were stained with DHE for ROS detection. ROS were markedly increased in the lungs of hyperoxia-exposed WT mice compared with air-exposed WT mice. This hyperoxia-induced increase in ROS was significantly attenuated in NOX1-deficient mice but not in NOX2-deficient mice (Figure 3A).
To detect the various cell types generating ROS, we stained lung sections from WT and NOX2- and NOX1-deficient mice with DHE and an anti–SP-C antibody, or with DHE and an anti-VWF antibody. In wild-type mice, type II epithelial and endothelial cells were positive for DHE staining during hyperoxia in comparison with air-exposed WT mice (respectively: Figure 3B, panels a and b and Figure 3C, panels a and b). Hyperoxia-induced ROS in epithelial type II and endothelial cells were markedly decreased in NOX1-deficient mice, but were not modified in NOX2-deficient mice compared with WT mice (respectively: Figure 3B, panels c and d, and Figure 3C, panels c and d).
Similar results were found for isolated type II AECs (Figures 4A and 4C) and primary MLECs (Figures 4B and 4D) derived from WT and NOX1-deficient mice exposed to hyperoxia. In contrast, primary type II AECs and MLECs isolated from NOX2-deficient mice produced an equal amount of ROS compared with cells isolated from WT mice during hyperoxia stimulation (Figures 4A–4D).
Interestingly, under normoxic conditions, lungs of NOX1-deficient mice exhibited a small, but significant increase in ROS generation as compared with WT mice, possibly because of a compensatory effect of ROS produced by other NOX enzymes or NOX-independent enzymes (Figure 3A; and see Figure E1 in the online supplement). Indeed, we analyzed the expression of NOX enzymes (NOX1, NOX2, and NOX4) in deficient mice and we did not find any up-regulation (Figure E1). This increase was not found in primary type II AECs and in primary MLECs isolated from NOX1-deficient mice (Figures 4C and 4D). These results indicate that NOX1 but not NOX2 is mainly responsible for ROS formation in lungs and in primary type II AECs and MLECs.
Because ROS contribute to oxidant-induced cell death in several animal models and in humans, we investigated whether NOX1 and NOX2 participate in alveolar cell death during hyperoxia through ROS generation. First, we examined in situ DNA strand breaks by fluorescence TUNEL staining. Hyperoxia increased significantly the number of TUNEL-positive cells in the lungs of WT mice and NOX2-deficient mice. By contrast, a much lower number of cells was positive when derived from NOX1-deficient mice under hyperoxia (Figures 5A and 5B). These results were supported by those obtained with primary type II AECs and MLECs isolated from hyperoxia-exposed WT mice and from NOX2- and NOX1-deficient mice (Figures 5C and 5D).
Zhang and colleagues reported that caspase-3 activation participated in oxidant-induced DNA fragmentation and cell death in MLE-12 cells (6). We therefore examined whether ROS-induced cell death in situ involved caspase-3 activation and subsequent PARP-1 cleavage. In hyperoxic lungs, there was a significant increase in the amount of cleaved caspase-3 (17 kD) and cleaved PARP-1 (85 kD) in WT mice, whereas in NOX1-deficient mice only small increases were observed (Figures 6A–6C). Indeed, the cleavage of procaspase-3 (32 kD) resulted in two fragments: a 19-kD fragment and a 17-kD fragment, with the 17-kD fragment being the active form of caspase-3 responsible for PARP-1 cleavage (31). We confirmed these results on lung sections from air- and hyperoxia-exposed WT and NOX1-deficient mice, using cleaved caspase-3 immunofluorescence staining (Figure 6D). In hyperoxia, there was a fivefold increase in caspase-3–positive cells compared with air in WT mice, whereas only a few positive cells were found in NOX1-deficient lungs: 5.6 ± 0.7% for WT mice versus 1.1 ± 0.2% for NOX1-deficient mice (the values are expressed as the mean percentage ± SEM; P < 0.001, n = 3). These data demonstrate that under hyperoxic conditions, NOX1-generated ROS but not NOX2-generated ROS mediate lung cell death, and NOX1 modulates the caspase-3–dependent apoptotic process.
To determine the mechanisms by which NOX1-derived ROS are responsible for lung alveolar cell death, we examined ROS-involved MAPK (JNK and ERK) pathways. Hyperoxia was associated with a significant increase in pJNKs and pERKs in WT mice as seen by Western blot. NOX1, but not NOX2, deficiency decreased ERK and JNK phosphorylation without modification of JNK and ERK protein levels (Figures 7A–7F).
To identify cell types involved in ERK and JNK activation in hyperoxia, lung sections from WT and NOX1-deficient mice were immunostained with an anti-pERK or pJNK antibody combined with an anti–SP-C antibody or an anti-VWF antibody. pERK staining was associated with type II pneumocytes and endothelial cell markers during hyperoxia (Figure 8A), whereas pJNK staining was detected mainly in endothelial cells (Figure 8B). The staining for pERKs and pJNKs was attenuated in lungs from NOX1-deficient mice (Figure 8A, panels b and c and panels e and f, respectively; and Figure 8B, panels b and c and panels e and f, respectively), confirming the results obtained by Western blotting (Figures 7D–7F). Some cells were positive for pJNK in hyperoxia that were neither SP-C– nor VWF-positive cells. Therefore, according to their localization and the absence of staining, we concluded that these cells were most probably alveolar macrophages (Figure 8B, panels c–f). These data demonstrate that NOX1 participates in oxidative stress–induced acute lung injury, involving ERK and JNK phosphorylation of pulmonary endothelial and epithelial type II cells.
NOX enzymes are considered critical contributors to a variety of diseases including cancers, cardiovascular disease, and diabetes. NOX1 has been shown to be expressed in a large number of tissues, including the lung, but its precise function has not been determined yet (20, 32). Several studies have demonstrated a role for NOX1 in the vascular system: NOX1 deficiency decreased blood pressure and provided protection from aortic dissection and aneurysm formation in response to angiotensin II (17, 18, 33).
In the present study, we found that NOX1, but not NOX2, participates in hyperoxia-induced lung injury through decreased ROS generation and cell death in alveolar epithelial and endothelial cells. Cell death prevention was associated with decreased phosphorylation of two main components of MAPK pathways: pERK and pJNK, two enzymes involved in the cell death pathway. In NOX1- and NOX2-deficient mice, exposure to hyperoxia leads to a huge neutrophil influx in BAL fluid, higher than in WT mice. Our results are supported by previous studies (34). Indeed, NOX2-deficient mice exposed to 48 hours of hyperoxia after acid aspiration, and p47phox- or NOX2-deficient mice infected with Escherichia coli, showed a greater amount of neutrophils compared with WT mice, without modification of lung injury (35). Moreover, a similar increased lung inflammatory response was noticed after influenza virus infection in NOX2-deficient mice, but in this case it was associated with reduced lung damage and improved lung function (36). Our data are also consistent with previous studies showing that neutrophil or macrophage depletion did not change lung damage in hyperoxic lung injury (37, 38). Indeed, hyperoxia-induced acute lung injury is one of the most established models of oxidative stress and alveolar cell death, which is not closely linked to the magnitude of the inflammatory response; in this way it is different from most cases of human acute respiratory distress syndrome (39). However, in another study, hyperoxia-exposed NOX2-deficient mice produced a lower number of neutrophils in BAL fluid compared with WT mice (22). In that study, mice exposed to acute hyperoxia developed lung fibrosis that was accompanied by a large inflammatory response. Because fibrosis is not a classical pattern of acute hyperoxic lesions, we cannot compare our data with these results. Our study suggests that protection provided by NOX1 deficiency during hyperoxia is not due to decreased ROS production by lung inflammatory cells, where NOX2 is preferentially expressed.
The results of this study demonstrate that hyperoxia-induced oxidative stress and tissue damage is, at least in part, mediated by NOX1. One of the reasons for the important role of NOX1 in hyperoxia might lie in the spatial localization of NOX1 in the lung. Primary pulmonary epithelial type II and lung endothelial cells expressed NOX1. Therefore, NOX1 is strategically located to be exposed to high oxygen concentration and to cause damage to the alveolocapillary barrier. A biochemical feature that distinguishes NOX1 from NOX2 might also play an important role. NOXO1, the organizer subunit of the NOX1 enzyme complex, lacks an autoinhibitory domain (40) and a basal NOX1 activity level has therefore been consistently reported (11). Thus, in contrast to NOX2, which is basically inactive in the absence of cell activation, we found that NOX1 is able to directly increase its ROS generation in response to the elevation of its substrate oxygen in murine alveolar epithelial type II and endothelial cells. In addition to increased substrate, that is, oxygen concentration, other mechanisms might contribute to the increased NOX1 activation under hyperoxic conditions, for example, kinase activation and/or activation of Rac GTPases (41).
NOX1 expression was increased in both epithelial and endothelial cells during hyperoxia. Specific absence of NOX1 decreased hyperoxia-induced ROS generation within the lung and specifically in pulmonary epithelial and endothelial cells. Decreased ROS production was significantly associated with less alveolar cell death as shown by DNA fragmentation markers, such as TUNEL staining and decreased active caspase-3 (17 kD) and PARP-1 cleavage. Also, DPI treatment, a nonspecific inhibitor of NOX enzymes and other electron transporters, decreased hyperoxia-induced ROS generation in human artery endothelial cells and oxidant-induced cell death in MLE-12 cells (6, 9). This cell death reduction was associated with decreased caspase-3 activation and PARP-1 cleavage in MLE-12 cells (6).
In the present study, hyperoxia activated both pERK and pJNK in mouse lungs as already reported (42) and particularly in pulmonary type II epithelial and endothelial cells. Interestingly, pJNK was found mostly in lung endothelial cells during hyperoxia, suggesting differential MAPK activation in vivo. The absence of NOX1 but not NOX2 decreased hyperoxia-induced JNK and ERK phosphorylation in alveolar cells. In accordance with these observations, treatment with DPI completely inhibited hyperoxia-induced ERK1/2 activation and subsequent cell death in alveolar epithelial cells (6, 43). Furthermore, the response to tumor necrosis factor–induced JNK activity and cell death was inhibited in alveolar epithelial cells transfected with NOX1-specific small interfering RNA (44, 45). It is interesting to note that JNK activation was similarly detectable in alveolar macrophages during hyperoxia in NOX1-deficient mice and in WT mice, thus attesting to the specificity of JNK activation by NOX1 in alveolar cells (46). The precise redox-sensible step involved in kinase activation in response to NOX-dependent ROS production is presently unknown. It may be due to inhibition of MAPK tyrosine phosphatase activity or activation of apoptosis signal-regulating kinase-1 (ASK1)-dependent or -independent pathways (44, 47, 48). It has been proposed that ROS, via the oxidation of thioredoxin, a redox regulatory protein, leads to ASK1 liberation and to tumor necrosis factor–induced JNK activation in MLE-12 cells (44). We also explored Akt phosphorylation (pAkt), because the instillation of the constitutively active form of Akt in mice before exposure to hyperoxia protected lung from oxidant-induced lung damage (49). Our results did not favor the hypothesis that a change in pAkt could mediate lung cell protection in NOX1-deficient mice (data not shown), but they strongly suggest that ROS generated by NOX1 participate specifically in the signaling of hyperoxia-induced ERK and JNK activation and thereby induce alveolar cell death in mouse lung.
Our data are the first direct demonstration that NOX1 activation promotes apoptosis in vivo and especially in the cells that comprise the alveolocapillary barrier. The results are not compatible with the hypothesis that ROS released by mitochondria are the primum movens of hyperoxic oxidative stress. We also found that NOX-derived ROS cause DNA strand breaks in vivo. However, it is not clear at this point whether NOX1-derived ROS are responsible for direct genotoxic stress or whether the DNA strand breaks are also due to caspase activation by redox-sensitive cell death pathways (e.g., JNK and ERK). Indeed, it is likely that caspase activation is not sufficient to fully account for the DNA strand breaks observed in response to hyperoxia because (1) in this study, we observed a smaller number of caspase-3–positive cells, as compared with TUNEL-positive cells; and (2) in previous studies we found that administration of pan-caspase inhibitor (Z-VAD) was not sufficient to rescue hyperoxia-induced cell death (3, 50). If direct genotoxic stress through NOX1-derived ROS is a contributor to hyperoxic lung damage, it is essential to inhibit ROS formation at a significantly upstream level. This approach could have an important impact on future therapeutic developments, potentially favoring the targeting of NOX1-specific inhibitors over inhibition of redox-sensitive cell death pathways in the treatment of acute lung injury.
The authors thank Philippe Henchoz, Camille Bron, Karim Hamad, Olivier Basset, Christian Vesin, and Angela Baia for technical assistance, and Karen Bedard and Olivier Preynat-Sauve for scientific help and discussion. The authors also express gratitude to Hal Chapman (Department of Medicine, UCSF) for helpful technical advice in alveolar type II epithelial cell isolation and stimulating scientific discussions.
Supported by Swiss National Research Foundation grant 3100A0-109339, by the Eagle Foundation and the Von Meissner Foundation (C.B.A.), and by NHLBI R01 HL51854 (M.A.M.).
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript.