Although TNFR1 is sufficient to induce sensitized apoptosis (in the presence of transcriptional or translational inhibitors) in primary lung mouse MVEC, the presence of both receptors is required for direct TNF-induced cell death (Lucas et al., 1998). While TNF-induced apoptosis in lung endothelium involves myosin light chain (MLC) phosphorylation, TNF-induced capillary barrier dysfunction does not (Petrache et al., 2001). Elevated plasma levels of soluble TNFRs are associated with morbidity and mortality in ALI patients (Parsons et al., 2005). Bronchoalveolar lavage fluid from early ARDS patients contains significantly elevated levels of both TNF and soluble TNF receptors, as compared to lavages from control patients investigated for chronic cough, hilar masses or pulmonary nodules (Hamacher et al., 2002).

TNF-TNFR2 interactions can be either destructive or preventive (Tartaglia et al., 1993; Lucas et al., 1997c; Wilson et al., 2007). On the one hand, TNF-TNFR2 signaling inhibits edema formation in a murine model of ventilator-induced lung injury, since mice lacking TNFR2 are more prone to the formation of edema, whereas mice not expressing TNFR1 are protected (Wilson et al., 2007). On the other hand, as indicated above, TNFR2 may increase TNFR1-mediated apoptosis upon impairing anti-apoptotic pathways (Tartaglia et al., 1993; Lucas et al., 1998; Li et al., 2002). As both TNF receptor types can mediate divergent signaling pathways, their relative expression in lung endothelial and alveolar epithelial cells can profoundly affect TNF signaling during ALI (Bertok et al., 2011).

Receptor interacting protein kinase 1(RIPK1)–dependent endothelial necroptosis underlies the TNF-induced systemic inflammatory response syndrome in mice (Zelic et al., 2018) and the blocking of connexin 43 hemichannels, which promote Ca²⁺ influx, with TAT-Gap19 protected mice against TNF-induced vascular leakage and mortality (Delvaeye et al., 2019). Laminar flow fosters pro-survival signaling, at least partially by impairing TNF-induced caspase-3 activity (Garin et al., 2007).

TNF not only decreases barrier function in ARDS by inducing cell death, but it can also affect junction protein status. Increased endothelial Ca²⁺-influx is the initial pivotal signal preceding pathways leading to contraction, including myosin light chain (MLC)-dependent mechanisms and MLC-independent microtubule rearrangement (Sandoval et al., 2001). TNF induces F-actin depolymerization in bovine pulmonary artery endothelial cell monolayers (Goldblum et al., 1993; Matsumura et al., 1997) to promote intracellular gap formation and barrier dysfunction. Across human pulmonary artery endothelial cells, TNF decreases trans-endothelial electrical resistance. This process occurs independently of MLC phosphorylation, which is catalyzed by either MLC kinase or Rho-kinase (Petrache et al., 2003). Rather, TNF causes microtubule destabilization and actin rearrangement in human pulmonary artery endothelial cells (Petrache et al., 2003). TNF-induced disassembly of the peripheral microtubule network involves activation of the p38 mitogen-activated protein kinase pathway in endothelial cells of human pulmonary arteries, which leads to an increase in permeability of endothelial cell layers (Petrache et al., 2003). Depolymerization of microtubules causes disassembly of adherens junction proteins they associate with, such as VE-cadherin, which increases permeability (Petrache et al., 2003). TNF can increase tyrosine phosphorylation of VE-cadherin (Angelini et al., 2006) and degrade the endothelial glycocalyx (Nieuwdorp et al., 2009). The Ca²⁺–dependent protein kinase C-α isoform can also play an important role in initiating endothelial cell contraction and disassembly of VE-cadherin–mediated cell–cell contacts (Petrache et al., 2003).

Tumor necrosis factor can also interact with other pathways promoting vascular leak. As such membrane-bound TNF (mTNF) and VEGF signaling pathways converge at the level of the secondary messenger, SAPK-2/p38, with continuous autocrine signaling by mTNF sensitizing endothelial cells to respond to VEGF by increasing their vascular permeability (Petrache et al., 2003).

Although pulmonary venous and capillary endothelial cells and are at times highly exposed to O₂-rich environments, they use the glycolytic pathway, rather than mitochondrial respiration as the main ATP-generating pathway in quiescent and proliferating states. They likely do so in order to maximize O₂ delivery to tissues and perivascular cells. This prevents the generation of high levels of reactive oxygen species in this cell compartment (Xiao et al., 2021). The endothelium, however, also provides the first line of defense against inflammatory or infectious insults in the vasculature. Upon inflammatory stimulation by TNF, otherwise quiescent endothelial cells are activated to produce significantly increased levels of pro-inflammatory molecules and NADPH oxidase-generated ROS (Gertzberg et al., 2004), and they moreover increase PFKFB3-mediated glycolysis (Cao et al., 2019). ROS can in turn combine with nitric oxide (NO) to generate barrier-disruptive peroxynitrite (Neumann et al., 2006). Although endothelial cells were shown to express NOX-1, -2, -4 and -5, especially NOX-2 was demonstrated to contribute to TNF-induced ROS generation (Deng et al., 2012; Drummond and Sobey, 2014).

Reduction of endothelial nitric oxide synthase (eNOS)-mediated NO generation was shown to promote pulmonary microvessel leakage in eNOS^–/– mice (Predescu et al., 2005), suggesting a barrier-protective effect of eNOS-derived NO in the pulmonary microvasculature (Di Lorenzo et al., 2013). Increased ROS generation was shown to impair the expression and activity of eNOS (Neumann et al., 2004) and to increase the activity of arginase 1, an enzyme competing with eNOS for the common substrate arginine (Romero et al., 2008; Chandra et al., 2012; Lucas et al., 2013). TNF was shown to be critical for arginase 1 induction in a murine model of ischemia and reperfusion, since TNF^–/– mice had significantly reduced endothelial arginase activity (Gao et al., 2007). All of these events involve increased endothelial oxidative stress and can disrupt the vascular integrity by reducing adherens and tight junction protein expression and/or by inducing endothelial cell death. ROS can moreover increase expression of adhesion molecules in endothelium and as such foster transendothelial migration of circulating leukocytes (Wang and Doerschuk, 2000). A stop of blood flow, as can occur in TNF-induced multiple organ failure, was moreover shown to activate a mechano-sensing machinery that generates ROS and that subsequently drives neutrophil influx (Noel et al., 2013). By contrast, laminar shear stress exerts a profound anti-apoptotic effect on endothelial cells treated with TNF, via the increased NO synthesis and post-transcriptional activation of eNOS (Cicha et al., 2009).

Apart from directly inducing barrier impairment, TNF can also indirectly induce hyperpermeability. TNFR1 activation increases neutrophil and monocyte adhesion to endothelium (Zhao et al., 2014), through upregulation of cell surface expression of endothelial cell adhesion molecules such as E-selectin, ICAM-1 and VCAM-1 in activated human pulmonary microvascular endothelial monolayers in vitro (Zhou et al., 2007; Su et al., 2008; Proudfoot et al., 2018). The induction of VCAM expression is associated with ROS generation, and endothelial NOX-2 is a major source of ROS associated with pulmonary inflammation (Frey et al., 2006). In patients and mice with critical limb ischemia, increased expression of the Cu²⁺ transport protein antioxidant-1 (Atox1) stimulates transcription of the NADPH oxidase organizer p47phox, thereby increasing VCAM-1/ICAM-1 expression and promoting monocyte adhesion in endothelial cells inflamed with TNF (Chen et al., 2015; Das et al., 2019).

The TNF-TNFR1/p38 MAPK/MK2 signaling axis induces secretion of pro-inflammatory chemokines, such as the neutrophil attractant IL-8 in human lung MVEC (Su et al., 2008). Moreover, TNF-induced activation of PI3Kγ in endothelial cells in turn induces PKC-ζ-dependent activation of NOX, leading to the generation of ROS and thereby degradation of the proteasome-dependent inhibitor of NF-κBα –IκBα- which further activates NF-κB-dependent inflammation (Frey et al., 2006).

TNF can promote thrombosis by inducing luminal endothelial expression of tissue factor. Tissue factor interacts with circulating coagulation factor VII and as such triggers the extrinsic coagulation pathway. This process involves NADPH oxidase activation (Clauss et al., 1996; Mechtcheriakova et al., 2001). Selective inhibitors of either TNFR1 or TNFR2-induced signaling pathways were shown to inhibit tissue factor expression (Clauss et al., 1996; Mechtcheriakova et al., 2001). However, TNFR1 is the predominant receptor responsible for the synergistic interaction between VEGF and TNF in endothelial tissue factor expression (Clauss et al., 1996; Mechtcheriakova et al., 2001).

Acute Respiratory Distress Syndrome develops in about 20% of COVID-19 patients with severe symptoms (Matthay et al., 2020; Du et al., 2021) and is influenced by age-, sex-, the presence of co-morbidities like diabetes, obesity and arterial hypertension, immunodeficiency and vaccination status. The development of COVID-19 associated ARDS is highly age-dependent and represents the main cause of COVID-19 related morbidity and mortality. As a matter of fact, COVID-19-related ARDS has increased the prevalence of ARDS in most highly industrialized countries by a factor of three (Matthay et al., 2020), leading to immense problems in terms of human resources, including excessive work load for physicians, nurses and other health care workers, with a profound impact on patients and their families, including social isolation and reduced patient care for other diseases. Severe COVID-19 is characterized by increased permeability of the alveolar-capillary barriers combined with diffuse alveolar damage, as shown in autopsies (Zhang et al., 2020). Human lung capillary endothelial cells express the main receptor for SARS-CoV2, i.e., angiotensin converting enzyme 2 (ACE-2) (Gheblawi et al., 2020), as well as of the cellular proteases, furin (Mayer et al., 2004) and TMPRSS2, necessary to mature the spike glycoprotein for subsequent receptor binding and membrane fusion (Hoffmann et al., 2020). The combination of these post-translational events causes endothelial cells to be vulnerable to SARS-CoV2. Some groups are detecting SARS-CoV2 within the pulmonary endothelium of COVID-19 patients (Ackermann et al., 2020; Varga et al., 2020) and in infected human blood vessel organoids (Monteil et al., 2020). Although therapeutic strategies to blunt host damage in COVID-19 mainly focus on the alveolar epithelium, COVID-19, particularly in the later complicated stages, is a disease of endothelial cells, particularly in the pulmonary microvasculature (Libby and Lüscher, 2020).

COVID-19-associated ARDS is accompanied by a significant imbalance in the cytokine/chemokine profile (Pedersen and Ho, 2020), with increased plasma levels of pro-inflammatory cytokines and chemokines (Pedersen and Ho, 2020). The severity of the inflammatory response, combined with T cell lymphopenia significantly contributes to pulmonary devastation and respiratory distress (Pedersen and Ho, 2020). An excessive inflammatory response during SARS-CoV-2 infection, which includes high levels of TNF, can provoke death of resident lung cells, which can further amplify barrier dysfunction. Elevated BAL and plasma levels of TNF are a hallmark manifestation of COVID-19, and can be associated with significant morbidity and mortality rates (Del Valle et al., 2020; Robinson et al., 2020). Increased soluble TNF levels can be caused by SARS-CoV2-induced activation of TACE (Haga et al., 2008; Atzeni et al., 2021). Mast cells activated by SARS-Cov-2 infection in respiratory tracts are also an important source of TNF (Kritas et al., 2020; Motta Junior et al., 2020) and together with the virus they can activate macrophages (Kritas et al., 2020). TNF is a major mediator of pulmonary edema and multi-organ failure in cases of severe COVID-19 (Eisenhut and Shin, 2020). TNF also promotes viral entry and dissemination since it induces adhesion molecule expression in endothelial cells and promotes VEGF generation, all of which contribute to increased permeability of the endothelial layer and thereby dissemination of SARS-CoV2 in patients with COVID-19 (Polidoro et al., 2020). Hyperpermeability of pulmonary blood vessels also results in improper accumulation of inflammatory monocytes, macrophages and neutrophils in the alveolar space (Channappanavar et al., 2016). Lung biopsies from patients who succumbed to severe COVID-19 show remarkable pathological injuries, with alveolar exudates, proliferation of alveolar epithelium, interstitial inflammation, and formation of hyaline membranes (Tan et al., 2020).

Activation of the coagulation pathway is another known feature of severe COVID-19 with potential development of disseminated intravascular coagulation (Teuwen et al., 2020). This is associated with pulmonary endothelial cell activation and dysfunction as cell death and the disruption of vascular endothelial integrity leads to exposure of the thrombogenic basement membrane and further in the activation of the coagulation cascade (Nachman and Rafii, 2008). In addition, IL-1β and TNF activate endothelial cells, which causes activation of coagulation cascade by expressing P-selectin, von Willebrand factor and fibrinogen, that are required for attaching platelets (Pober and Sessa, 2007).

COVID-19 endothelialitis is associated with platelet aggregation and thrombocoagulatory and fibrinolysis-inhibitory (D’Agnillo et al., 2021) features that can obstruct vessels, as manifested by the important frequency of thromboses, including pulmonary thromboembolism. This has been shown particularly in the pulmonary microvasculature, especially in vessels with a cross-sectional area between 1.25 and 5 mm² (Libby and Lüscher, 2020; Morris et al., 2021), containing highly activated and functionally distinct neutrophils, which generate neutrophil extracellular traps (NETs) (Leppkes et al., 2020; Zuo et al., 2020; Panda et al., 2021), which may stimulate alternative treatment strategies besides corticosteroids and standard care (Panda et al., 2021). One has to question whether virus-associated endothelial cell injury, quite frequently co-localized with NETs, or rather the collateral of other inflammatory processes in plasma, represents the key step to neutrophil activation. The high pro-coagulation context in COVID-19 in part motivates therapeutic anticoagulation approaches especially with heparins in hospitalized patients who are not critically ill (Leppkes et al., 2020; Houston et al., 2020). Such patients seem to profit and seem less in danger of anticoagulation than sicker and frailer poly-monitorized ventilated ARDS patients, in whom both risk and benefit need to be more closely examined. To circumvent progressive respiratory failure, prone positioning of the patient, which has high preventative and therapeutic value, is recommended since it improves ventilation and perfusion mismatch (Gattinoni et al., 2020). Conceptually, the high frequency of the above-mentioned pulmonary embolism, including the site of peripheral arterial large vessel occlusions like coronary, cerebral, limb, and other arterial occlusions, may have a similar pathogenetic background, which may also be the case for COVID-19 pneumonia and makes it a more privileged respiratory disease, as compared to non-COVID ARDS in terms of prognosis.

To counteract excessive deleterious TNF signaling leading to tissue damage, barrier dysfunction and hyperinflammation, several types of molecular-based therapeutic approaches have been developed over the past decades (Fischer et al., 2020). These include monoclonal antibodies such as infliximab (Remicade), adalimumab (Humira), certolizumab pegol (Cimzia), golimumab (Simponi) and soluble TNF receptor constructs, such as etanercept (Enbrel) (Russell et al., 2020). TNF inhibitors may directly compete with the TNF ligand, or deactivate its receptors or the enzyme generating the soluble TNF ligand, i.e., TACE, resulting in inhibition of downstream signaling pathways leading to cell death or inflammation. Figure 3 illustrates some of the different targets associated with TNF and TNFRs that can be inhibited by natural or pharmacological TNF blockers. In addition, data obtained from various studies and clinical trials on the impact of natural or pharmaceutically designed TNF blockers are summarized in Tables 1A,B.

Despite the involvement of a vast variety of pro-inflammatory mediators, exclusive blockage of TNF seems to result in better clinical outcomes in a broad range of diseases associated with inflammation and autoimmunity and clinical trials assessing TNF inhibitors for treatment of ARDS in severe COVID-19 patients have been proposed (Feldmann et al., 2020). Drugs with anti-inflammatory and/or anti-viral effects can be considered as add-on therapy against acute lung injury- (Gallelli et al., 2020) and COVID-19-associated hyper-inflammation (Robinson et al., 2020). Indeed, genetic or pharmacological inhibition of TNF and TNFRs abrogates production of TNF in pleural exudates and lung tissues and reduces infiltration of inflammatory cells into the pleural cavity and lung tissues, alveolar-capillary barrier dysfunction and resident lung cell apoptosis (Mazzon and Cuzzocrea, 2007). TNF neutralization has shown protection against SARS-CoV2 infection in animal models (Jamilloux et al., 2020).

As a note of caution, the observation of higher rates of, sometimes fatal, bacterial infection in patients with rheumatoid arthritis and psoriasis treated with TNF neutralizing antibodies and soluble receptors (Keane et al., 2001) clearly documents the danger of chronically suppressing these sometimes crucial bactericidal pathways. Taken together, these findings demonstrate limitations of chronically using TNF neutralizing agents that blunt binding of the ligand to its receptors, since this approach can inhibit both deleterious and beneficial TNF-TNFR signaling. The search for alternative treatments that selectively impair TNF-mediated vascular dysfunction, AFC impairment and hyper-inflammation in cells of the alveolar-capillary compartment, without impairing the cytokine’s beneficial actions in host response to infection, is therefore therapeutically significant.

Alveolar fluid clearance is essential to keep the alveolar space relatively dry, to assure efficient gas exchange. AFC is mediated by vectorial Na⁺ transport through the apically expressed epithelial sodium channel (ENaC) and basolateral Na⁺-K⁺-ATPase in type I and II alveolar epithelial cells (Matthay et al., 2005; Vadász et al., 2007). The amiloride-sensitive ENaC in its native configuration consists of three subunits of α, β, and γ and controls the overall rate of transapical Na⁺ transport (Dagenais et al., 2004). A fourth δ subunit was also identified and can replace the α subunit to form an alternative active ENaC configuration (Ji et al., 2012). Na⁺ absorption by ENaC is the critical driving force of AFC at birth and is crucially involved in pulmonary edema clearance in adulthood (Dagenais et al., 2004). Hence, mice lacking ENaC-α die from alveolar flooding at birth (Hummler et al., 1996).

Epithelial sodium channel activity is regulated by subunit maturation upon cleavage of the extracellular loops of the α and γ subunits by proteases such as furin and prostasin (Kleyman and Eaton, 2020), by its cell surface expression N and by its open probability time Po, the latter of which is at least partially controlled by complex formation of ENaC with the protein MARCKS and with PIP₂ (Alli et al., 2012, 2015; Archer et al., 2020).

Conflicting study outcomes are found with regard to the effect of TNF on AFC. Whereas TNF generation is increased in lavage from acute ARDS patients (Hamacher et al., 2002), whose AFC capacity inversely correlates with mortality (Ware and Matthay, 2001), a TNF-dependent increase in AFC is found in a number of animal models, three of which are presented in more detail (Rezaiguia et al., 1997; Börjesson et al., 2000; Tillie-Leblond et al., 2002; Hamacher et al., 2018). Further studies show that TNF can either stimulate (Fukuda et al., 2001) or inhibit ENaC in alveolar epithelial cells (Johnson et al., 2002, 2006; Dagenais et al., 2004; Eaton et al., 2004) and that, accordingly, the cytokine has a substantial impact on the capacity of alveolar epithelial cells to transport Na⁺ (Dagenais et al., 2004). In the following paragraphs we will discuss the opposing actions of TNF-TNFR1 signaling and activities of the lectin-like domain of TNF on ENaC.

The inhibitory effect of TNF on ENaC is mainly mediated by TNFR1 (Dagenais et al., 2004; Braun et al., 2005). TNF decreases the mRNA expression levels of α, β, and γ subunits of ENaC in alveolar epithelial cells isolated from male Sprague-Dawley rats (Dagenais et al., 2004), which correlates with a significant reduction in amiloride-sensitive current in these cells (Dagenais et al., 2004). TNF moreover affects the stability of ENaC mRNA (Dagenais et al., 2004). In addition, TNF downregulates ENaC-α mRNA involving a complex process that requires the transcription of multiple genes (Dagenais et al., 2004). Along with the abovementioned results, direct exposure of rat type II alveolar epithelial cells to TNF inhibits mRNA and protein expression of the α- and γ-, but not of the β subunit of ENaC (Yamagata et al., 2009) and impairs ENaC function, as revealed by reduced amiloride-sensitive currents in these cells (Yamagata et al., 2009). These results document the substantial negative influence TNF can exert on the capacity of alveolar epithelial cells to transport sodium (Dagenais et al., 2004).

One of the mechanisms through which TNF can impair ENaC activity is through the TNFR1-mediated activation of acid sphingomyelinase (ASMase), an enzyme located in lysosomes and in the extracellular space, which catalyzes the conversion of sphingomyelin, a major component of membranes, into ceramide and phosphocholine (Gorelik et al., 2016). Low levels of ASMase maintain an equilibrium within the alveolar space between ceramides and sphingomyelin substrate (Ryan et al., 2003), but during inflammation this balance can be altered, resulting in increased generation of injurious sphingolipids, that can in turn accelerate pulmonary edema formation (Schissel et al., 1996; Haimovitz-Friedman et al., 1997; Ryan et al., 2003). Ceramide is a crucial mediator of TNF/TNFR1-induced inhibition of ENaC, via a pathway associated with PKC-dependent externalization of phosphatidyl serine (Bao et al., 2007).

Apart from inhibiting ENaC, TNF/TNFR1 signaling can also affect mRNA and/or protein expression of other channels involved in AFC. As such, TNF impairs expression of the major water channel aquaporin 5, expressed in alveolar, tracheal, and upper bronchial epithelium, through a mechanism involving activated NF-κB (Towne et al., 2000, 2001). Moreover, TNF downregulates the steady-state mRNA level of the cystic fibrosis transmembrane conductance regulator (CFTR) in HT-29 cells (Nakamura et al., 1992). Depending on the direction of the chloride transport, CFTR can either activate (inward) or inhibit (outward) ENaC activity. Similar to the impact of TNF on ENaC, its inhibitory action on CFTR occurs at the post-transcriptional level in a dose- and time-dependent manner (Nakamura et al., 1992), by reducing CFTR mRNA half-life (Nakamura et al., 1992).

In a Pseudomonas aeruginosa (P. aeruginosa)-induced pneumonia model in rats, Rezaiguia et al., measured alveolar protein and liquid clearance, 24 h after intratracheal instillation of bacteria, by instilling 5% bovine albumin solution with ¹²⁵I-labeled human albumin into the airspaces (Rezaiguia et al., 1997). The concentration of unlabeled and labelled protein in the distal airspaces over 1h is used as an index of net AFC (Rezaiguia et al., 1997). Surprisingly, the investigators detect a P. aeruginosa pneumonia-induced increase in AFC over 1h by about 48% in experiments with blood flow, and by 43% in those without blood flow, compared with control groups (Rezaiguia et al., 1997). In both experiments, this AFC increase is inhibited by amiloride, which blunts ENaC activity. As propranolol has no inhibitory effect on AFC, a catecholamine-dependent mechanism can be excluded (Rezaiguia et al., 1997). Intratracheal instillation of an anti-TNF neutralizing antibody into the lungs 5 min before bacterial instillation is shown to prevent the increase in AFC in rats with pneumonia (Rezaiguia et al., 1997) and instillation of 5 mg TNF in normal rats increased AFC by 43% over 1h compared to control rats receiving vehicle (Rezaiguia et al., 1997), an effect not blunted by propranolol. As such, TNF at early time points following G^– pneumonia can upregulate net AFC, partly by a TNF-dependent mechanism (Rezaiguia et al., 1997). This study thus provides evidence of a positive role of TNF in AFC, in a catecholamine-independent manner.

Studies by Börjesson et al. (2000) in ventilated, anesthetized rats with 45 min intestinal ischemia followed by 3 h reperfusion show development of remote organ injury, including the lungs. A 76% increase in AFC as compared to controls, measured as the increase in alveolar protein concentration over 45 min, is at least partially mediated by a TNF-dependent mechanism, since systemic administration of a neutralizing polyclonal anti-TNF antibody before induction of intestinal ischemia fully inhibits the rise in AFC. Similar to the results above (Rezaiguia et al., 1997), the stimulated AFC after intestinal ischemia-reperfusion cannot be inhibited by propranolol and is therefore assumed to be catecholamine-independent (Börjesson et al., 2000). Intracellular cAMP levels are not increased in this model and amiloride inhibits similar fractions of AFC in animals subjected to ischemia-reperfusion or in control animals (Börjesson et al., 2000). In conclusion, the rat intestinal ischemia-reperfusion model leads to an AFC increase that is TNF-dependent.

Evaluation of AFC and epithelial and endothelial permeability in a model of chronic allergic inflammation in Brown-Norway rats challenged with ovalbumin by Tillie-Leblond et al., demonstrates lung sections with inflammatory infiltrates associated with perivascular and peribronchiolar edema in the ovalbumin group (Tillie-Leblond et al., 2002). To measure AFC in this study, a 5% bovine albumin solution with ¹²⁵I-labeled human albumin is instilled into their air spaces and alveolar-capillary barrier permeability is evaluated in another group of animals by intravenous injection of 1 μCi of ¹³¹I-labeled albumin (Tillie-Leblond et al., 2002). Endothelial permeability in this study significantly increased in the ovalbumin group but the final-to-initial protein ratio reaches significantly higher levels in the ovalbumin group, as compared to the control group (Tillie-Leblond et al., 2002). Administration of neutralizing rat anti-TNF antibodies in the instillation significantly decreases this ratio. As such, also in this rat model of severe bronchial allergic inflammation associated with increased endothelial and epithelial permeability, a TNF-dependent increase in AFC is observed.

These three in vivo studies of G- pneumonia, intestinal ischemia and reperfusion or ovalbumin-induced inflammation give clear evidence of an increase in AFC involving a crucial role of TNF (Hamacher et al., 2018).

One of the explanations for the observed AFC-promoting actions of TNF discussed above can come from either positive actions of TNFR2 signaling or from a role for the lectin-like domain of TNF. Indeed, in sharp contrast to the TNFR1 binding site, the lectin-like domain of TNF, mimicked by the TIP peptide (a.k.a. Solnatide, AP301), can activate amiloride-sensitive Na⁺ uptake in alveolar epithelial cells, as well as in lung and glomerular MVEC, which also express all ENaC subunits (Hribar et al., 1999; Elia et al., 2003; Braun et al., 2005; Madaio et al., 2019). This effect of the lectin-like domain of TNF occurs even in cells lacking both TNF receptors (Hribar et al., 1999) thus clearly documenting that the TIP peptide binds to partners different from the TNF receptors. Actually, TIP peptide binds to the α subunit of ENaC and as such directly increases the channel’s activity. This occurs by affecting both the open probability (Po) and the cell surface (N) expression of ENaC (Czikora et al., 2014; Shabbir et al., 2015; Lucas et al., 2016). Treatment of cells with N-glycosidase F, which removes N-linked glycosylation from ENaC subunits, completely abrogates channel activation by the TIP peptide (Czikora et al., 2014). This interaction mainly affects the promotion of ENaC’s cell surface expression and involves the Asn511 residue in the extracellular loop of ENaC-α (Shabbir et al., 2015). The effects of the peptide on ENaC open probability rather involve a different interaction site within the second transmembrane domain of ENaC-α in which Val567 and Glu568 are crucial (Lucas et al., 2016). The TIP peptide even activates ENaC in the presence of bacterial toxins like pneumococcal pneumolysin, a pore-forming toxin which impairs both the Po (Lucas et al., 2012a) and N of ENaC (Lucas et al., 2016), both in vitro and in vivo (Czikora et al., 2014; Hamacher et al., 2018).

Despite of the negative effects of TNFR1 on ENaC, the lower affinity of TNF for oligosaccharides in glycoproteins than for its membrane-associated TNF receptors (Hession et al., 1987; Sherblom et al., 1988) and the much higher expression level of TNF receptors than of ENaC in alveolar epithelial cells, an intriguing observation is that inhibition of TNF with a neutralizing antibody worsened edema formation in a rat pneumonia model (Rezaiguia et al., 1997). An important question to ask is under what conditions can the beneficial activity of the lectin-like domain of TNF on ENaC predominate the deleterious one exerted by TNF-TNFR1 signaling? In an isolated perfused rat flooded lung model, which mimics cardiogenic (hydrostatic) edema, human TNF significantly decreases lung liquid clearance. Whereas the neutralizing anti-TNF monoclonal antibody Infliximab restores basal liquid clearance in this model, complex formation of TNF with its soluble TNF receptor 1 increases fluid clearance above basal levels, comparable to treatment of the lungs with TIP peptide (Braun et al., 2005). Whereas Infliximab likely interferes with both the receptor binding sites and the lectin-like domain of TNF, the soluble TNFR1 construct only blunts the former (Lucas et al., 1994; Braun et al., 2005). As such, it is tempting to hypothesize that the lectin-like domain of TNF can play a physiologically relevant role especially when present in complexes between TNF and its soluble TNF receptors, which blunt the receptor binding sites but leave the lectin-like domain active. This condition can occur in pneumonia, as such providing a potential explanation for the observed positive effects of TNF in a rat model of severe bacterial pneumonia (Rezaiguia et al., 1997). Triple mutant TNF mice expressing a mutant TNF lacking a functional lectin-like domain exhibit increased lung injury upon pneumolysin instillation as compared to wild type mice (Czikora et al., 2014). These findings indicate that the lectin-like domain can exert its protective role within the context of the entire TNF molecule and suggests that it represents an intramolecular negative feedback mechanism to keep a brake on TNF’s potent inflammatory activities.

Important progress in the understanding of the biological role of the lectin-like domain of TNF comes from in vivo pneumonia and ARDS models evaluating the potential therapeutic potential of the TIP peptide. As shown in Figure 4, both ENaC (in its native configuration α, β, γ) and the non-selective cation channel (NSC, consisting of an ASIC1a and an ENaC-α subunit), can be expressed in lung MVEC (Czikora et al., 2017). TIP peptide, which binds to ENaC-α, can strengthen barrier function in cells treated with the pore-forming bacterial toxins listeriolysin-O (LLO, from Listeria monocytogenes) and pneumolysin (PLY, from Streptococcus pneumoniae). Both LLO and PLY induce Ca²⁺-influx, which in turn activates Ca²⁺-dependent PKC isozymes, such as PKC-α, involved in capillary barrier dysfunction (Xiong et al., 2010; Lucas et al., 2012b). Both LLO and PLY impair ENaC activity (Lucas et al., 2012a; Yang G. et al., 2018). Our results show a profound protective effect of TIP peptide toward both LLO and PLY-induced barrier dysfunction, assessed by electrical cell substrate impedance sensing (ECIS), in human lung microvascular endothelial cell (MVEC) monolayers in vitro and in mice in vivo, the latter assessed by Evans Blue Dye incorporation (Xiong et al., 2010; Lucas et al., 2012b). A crucial role for ENaC-α in this process is documented by the significantly elevated sensitivity to PLY-induced barrier dysfunction in human lung MVEC in which ENaC-α is depleted by specific siRNA. Moreover, the protective effect of both the TIP peptide and of the NSC activator MitTx is lost under these conditions. TIP peptide blunts PLY-induced phosphorylation of calmodulin-dependent kinase II (CaMKII), as well as of its substrate filamin A (FLN-A) (Czikora et al., 2017). Non-phosphorylated FLN-A both promotes ENaC open probability and blunts stress fiber formation, thus providing a mechanism and a link for the ENaC activating and barrier protective role of the lectin-like domain of TNF.

Transplanted lungs often experience severe alveolar epithelial and endothelial dysfunction immediately following the procedure, due to ischemia-reperfusion injury. In the rat left-sided lung transplantation model of severe acute lung injury due to 20 h of cold ischemia and 24 h of reimplantation response, TIP peptide given intratracheally immediately before reimplantation into the left lung increases the left lung PaO₂/FiO₂ (P/F) ratio from about 70 mm Hg -indicating a condition similar to severe ARDS- to more than 400 mm Hg, which rather corresponds to a mild hypoxia-induced lung injury where a patient can typically walk up even stairs (Hamacher et al., 2010, 2018). Patient with a similar drop in P/F ratio as the one observed in the rat transplantation study group not treated with TIP peptide need to be tracheally intubated and are completely dependent on both the ventilator’s respiratory work and on a high fraction of oxygen supplementation. When applied together with the pharmacological ENaC inhibitor amiloride, the protective effect of the TIP peptide is abrogated. Complex formation of TIP peptide with N, N′-diacetylchitobiose, a selective oligosaccharide inhibitor of the lectin-like domain of TNF, or use of an inactive mutant TIP peptide, in which three crucial residues (1 Threonine and 2 glutamic acids) are changed for alanine also fail to restore oxygenation. These data strongly suggest that the significant protective effect of the TIP peptide in a severe ischemia-reperfusion lung injury model involves the lectin-like activity of TNF and is due to the activation of Na⁺ uptake via ENaC (Hamacher et al., 2010). The reduced infiltration of neutrophils in the bronchoalveolar lavage fluid, as well as the reduced generation of reactive oxygen species (ROS, discussed in the next paragraph), further underlines that the lectin-like domain of TNF can modify important inflammatory pathways in lung ischemia-reperfusion injury.

One of the main reasons for endothelial dysfunction following lung transplantation is the alteration in blood flow, which results in shear stress that can be sensed by a mechanosensitive complex -the mechanosome- in the endothelial cell membrane. The components of this mechanosome include caveolae, the adhesion molecule PECAM, vascular endothelial growth factor receptor 2 (VEGFR2) and the adherens junction protein VE-cadherin (Chatterjee and Fisher, 2014; Chatterjee et al., 2015). The shear signal can be transduced by components of the cell surface membrane, resulting in the activation of NADPH oxidases and the generation of ROS. In a rat iso-transplantation model, a significantly increased ROS generation occurs in the transplanted lung 24h after the procedure, but this is prevented in by pretreatment of the transplanted lungs with TIP peptide, but not with an inactive mutant peptide. This ROS-reducing activity of the TIP peptide correlates with by a significantly reduced infiltration of neutrophils in the bronchoalveolar lavage fluid (Hamacher et al., 2010). Both of these effects can be blunted by the ENaC inhibitor amiloride. TIP peptide also blunts ROS generation in vitro in ovine pulmonary artery endothelial cells undergoing hypoxia-reoxygenation. The results from these preclinical studies suggest that the TIP peptide is a therapeutic candidate to treat ischemia-reperfusion injury associated with lung transplantation, since it can blunt both oxidative stress and inflammation in endothelium.

Another study documenting the anti-inflammatory capacity of the TIP peptide comes from a study using a rat model of high-altitude pulmonary edema (HAPE). In this study, treatment with TIP peptide (a.k.a. Solnatide, AP301) reduces leukocyte infiltration and pro-inflammatory cytokine levels in lavage, while improving occludin expression in alveolar epithelial cells and reducing edema formation (Zhou et al., 2017).

The TIP peptide also has an anti-inflammatory action in other organs than the lungs, as shown in a mouse glomerulonephritis model, in which animals were treated with nephrotoxic serum. In this model, local immune deposits in the glomeruli generate TNF, which activates pro-inflammatory pathways in both glomerular endothelial cells and podocytes. Intraperitoneal treatment with TIP peptide in this model significantly reduces inflammation (reduced systemic and local pro-inflammatory cytokine generation, and reduced renal leukocyte infiltration), as well as proteinuria and blood urea nitrogen (Madaio et al., 2019). TIP peptide also increases PGE₂ generation in glomerular endothelial cells, which in turn promotes eNOS-dependent NO generation and barrier function (Figure 4). Despite the capacity of TIP peptide to activate ENaC, it does not increase mean arterial blood pressure in mice, possibly because it is captured in the loops of Henle by Tamm-Horsfall protein (Hession et al., 1987; Sherblom et al., 1988) before it can reach the distal tubules of the kidney. Figure 5 and Table 2 give an overview of the divergent actions of the TNF receptor binding sites versus the lectin-like domain in models of acute lung injury.