Important findings on the pathogenic potential of heme have been reported in experimental mouse models of SCD and malaria. Although of different origin, pathologies in both disorders are primarily linked to large amounts of hemoproteins released from damaged RBCs. SCD is a genetic disorder caused by an amino acid exchange in the Hb β-chain, which leads to the production of abnormally shaped sickle cells prone to intravascular hemolysis (Ingram, 1957). Experimental SCD mouse models seem to exhibit typical signs of vascular inflammation (Belcher et al., 2003). Furthermore, as indicated by leukocyte infiltration and thrombosis, heme toxicity is likely associated with phenotypical alterations of the vascular endothelium (Belcher et al., 2014; Keleku-Lukwete et al., 2015). Independently, heme was responsible for pro-inflammatory M1 polarization of macrophages (Vinchi et al., 2016), formation of neutrophil extracellular traps (Chen et al., 2014), and triggering of acute chest syndrome (Ghosh et al., 2013) in SCD mice. In malaria, which is a protozoan disease, hemolysis is caused by infection with and replication of Plasmodium in RBCs (Miller et al., 2002). In mouse models of experimental malaria release of extracellular Hb and heme from damaged RBCs also had major pro-oxidant and pro-inflammatory effects. Hemolysis-derived heme was directly involved in inflammatory pathologies of experimental murine cerebral and non-cerebral malaria (Pamplona et al., 2007; Seixas et al., 2009).

CNS hemorrhage occurring spontaneously, like in subarachnoid hemorrhage and stroke or because of traumatic injuries, can be closely associated with the toxicity of cell-free Hb and heme. In a feline model exposure to cell-free Hb caused marked pro-oxidant damage of nerve cells in vivo, which was primarily due to lipid peroxidation (Sadrzadeh et al., 1987). Similar findings have been reported in an experimental rabbit model of preterm pup intraventricular hemorrhage, in which Hb and heme led to inflammatory changes and cell death in affected tissues (Gram et al., 2014). Furthermore, heme toxicity played an important role in the pathogenesis of brain injury in a mouse model of intracerebral injection of cell-free Hb (Ma et al., 2016).

Sepsis and its more complicated manifestations, severe sepsis and septic shock, are characterized by an excessive systemic inflammatory response to various acute injuries (Bone, 1991; Gotts and Matthay, 2016). The pathophysiology of this complex disorder is not understood in detail (Angus and van der Poll, 2013). Recently, cell-free Hb and heme have been shown to be involved in the pathogenesis of severe sepsis in a mouse model of cecal ligation and puncture polymicrobial sepsis in vivo (Larsen et al., 2010). Similarly, pro-inflammatory effects of extracellular Hb and heme from degraded erythrocytes worsened the survival rate in an experimental rat model of E. coli-mediated sepsis (Griffiths et al., 1995).

Numerous reports have associated transfusions of packed RBCs after prolonged storage with increased morbidity and mortality in trauma-induced hemorrhage among other conditions. Detrimental effects of stored RBCs have been associated with the so-called “storage lesion,” which is characterized by RBC alterations including loss of metabolites, decreased cell volume with accompanying formation of echinocytes and release of free Hb due to hemolysis (Lelubre et al., 2009). The storage lesion has been directly linked to the toxicity of cell-free Hb and heme in a model of guinea pigs, in which transfusion of senescent RBCs was more harmful if compared with fresh RBC preparations (Baek et al., 2012). Similarly, others have demonstrated that heme toxicity due to RBC storage lesion markedly aggravated the outcome in two independent mouse models of trauma-induced hemorrhage (Stapley et al., 2015; Graw et al., 2016). Finally, transfusion of senescent RBCs aggravated inflammation and worsened outcome in a canine model of infectious pneumonia (Wang et al., 2012a).

Heme toxicity is also involved in the pathogenesis of diseases in solid organs such as kidney. For example, experimental rhabdomyolysis, in which large amounts of intracellular hemoproteins such as myoglobin are released, cause heme-dependent acute kidney injury (AKI) (Nath et al., 1992). Moreover, the detrimental pro-oxidant effects of hemoproteins were shown to be directly involved in kidney cell damage in vitro and in vivo (Nath et al., 1995). More recent studies in mouse models have extended these earlier findings by demonstrating that the heme degradation product iron and the iron-sequestering protein ferritin are critically associated with heme-dependent renal injury (Zarjou et al., 2013; Bolisetty et al., 2015). Comprehensive overviews on the pathophysiology of heme toxicity in kidney diseases have been previously given (Tracz et al., 2007; Lever et al., 2016). As to cardiac disorders increased levels of heme seem to aggravate ischemia-reperfusion injury in experimental murine heart disease. This novel finding suggests an important role for heme in the pathogenesis of ischemic cardiomyopathy (Sawicki et al., 2015). Remarkably, heme toxicity was also critically involved in acute lung injury due to bromine inhalation in a mouse model (Aggarwal et al., 2016).

Patients with hemolytic disorders such as SCD exhibit increased serum levels of heme (Muller-Eberhard et al., 1968) and develop acute and/or chronic manifestations of heme toxicity (Nath and Katusic, 2012). Remarkably, in SCD patients heme-carrying RBC membrane microparticles were responsible for endothelial cell damage (Camus et al., 2015) and heme specifically affected T cell polarization via interaction with CD16+ monocytes (Zhong et al., 2014). A direct pathogenic role for heme toxicity has also been demonstrated in human malaria. Heme markedly down-regulated prostaglandin and transforming growth factor-β production in malaria (Andrade et al., 2010). Moreover, increased concentrations of heme were associated with higher susceptibility to malaria (Mendonca et al., 2012) and disease outcome was linked with systemic levels of extracellular heme in these patients (Elphinstone et al., 2016).

Similar to findings in mouse models, heme appears to play a role in the pathogenesis of severe sepsis in humans as indicated by lower serum levels of Hx, which decrease due to increased levels of heme (Larsen et al., 2010). Accordingly, poor outcome in sepsis patients has been associated with decreased serum levels of Hp and Hx (Janz et al., 2013; Lin et al., 2015). In the context of sepsis it is important to note that mice exhibit markedly higher tolerance to endotoxin relative to humans (Schaedler and Dubos, 1961).

Transfusion of packed RBCs after prolonged storage has been associated with an increased risk of death in critically ill patients (Wang et al., 2012b). Therefore, the potential risk of transfusing senescent RBCs needs to be addressed by prospective controlled studies to clarify this issue in more detail.

Hemolysis-derived heme may be indirectly involved in the pathogenesis of atherosclerosis. The oxidation of lipoproteins and other plasma proteins by pro-oxidant iron from cell-free Hb and heme plays an important pathophysiological role for the complex sequence of vascular events that cause atherosclerosis (Jeney et al., 2002, 2014; Nagy et al., 2010). In clinical studies on chronic ischemic cardiomyopathy increased levels of heme have been observed in cardiac biopsies from patients with failing hearts (Sawicki et al., 2015). Independently, up-regulation of heme levels have been associated with a worse clinical outcome in patients with heart failure (Khechaduri et al., 2013). Interestingly, heme toxicity has recently been implicated in the pathogenesis of heart failure via affecting the contractile function of cardiomyocytes (Alvarado et al., 2015).

A recent clinical study of patients with atypical hemolytic uremic syndrome (a rare thrombotic microangiopathy primarily observed in the kidney) identified heme as a critical secondary hit that triggers the clinical manifestation of this disorder. Specifically, heme in plasma from patients with atypical hemolytic uremic syndrome activated the alternative pathway of the complement cascade, which in turn caused endothelial cell activation (Frimat et al., 2013). These findings suggest that heme may have similar effects in the pathophysiology of other vascular inflammatory diseases.

The plasma protein Hx binds non-covalently heme with the highest affinity of any known protein (K_D 10⁻¹⁴) (Muller-Eberhard and Cleve, 1963; Muller Eberhard, 1970; Tolosano and Altruda, 2002) via its characteristic heme-binding pocket (Paoli et al., 1999). The major function of Hx appears to be neutralization and scavenging of excess free heme from the circulation. Up-take of heme-Hx complexes in the liver (Potter et al., 1993) is mediated via the scavenger receptor low-density lipoprotein receptor-related protein-1 (LRP1, synonymous with CD91) (Hvidberg et al., 2005; Vercellotti et al., 2016). Hx belongs to the acute-phase reactants, which include a number of plasma proteins such as C-reactive protein, α2-macroglobulin and α1-antitrypsin that are up-regulated in the liver as part of a systemic inflammatory response (Heinrich et al., 1990; Baumann and Gauldie, 1994). Hx is induced during the acute-phase response in rodents, but not in human, which might be due to evolutionary differences in rodent and human Hx gene promoters (Heinrich et al., 1990; Poli et al., 1990; Immenschuh et al., 1994). This species-specific difference of Hx gene regulation in rodents and humans correlates with the recently reported findings that Hx is up-regulated during sepsis in mice, but down-regulated in humans (Lin et al., 2015). Hx knockout mice exhibit a normal phenotype in non-challenged conditions, but are afflicted with heme-mediated renal and hepatic damage in conditions of experimental hemolysis (Tolosano et al., 1999; Vinchi et al., 2008). Remarkably, heme-mediated pathologies are aggravated in Hx/Hp double knockout mice (Tolosano et al., 2002) suggesting that these two plasma proteins represent a sequential protection system against the detrimental effects of hemolysis (Deuel et al., 2015; Smith and McCulloh, 2015; Figure 2).

In addition to Hx, other HBPs are likely to be involved in the regulation of systemic extracellular and also intracellular homeostasis of heme and may counteract its pro-oxidant effects. The functional significance of most known HBPs for neutralization and transport of heme is only incompletely understood. However, it has been pointed out that the specific heme-protein interactions of a given HBP determine its protective potential against the pro-oxidant effects of heme, respectively (Vincent et al., 1988; Vincent, 1989). Another extracellular HBP with major physiological significance is albumin, which binds heme with markedly lower affinity than Hx (K_D 1.2 × 10⁻⁸) (Little and Neilands, 1960), but exhibits markedly higher plasma concentrations relative to Hx (Adams and Berman, 1980) (Table 1). Other known extracellular plasma HBPs are α1-microglobulin (Allhorn et al., 2002) and α1-antitrypsin (Karnaukhova et al., 2012) (Table 1). Intracellular binding of heme by HBPs may not only protect against its potential pro-oxidant toxicity, but may be also involved in trafficking of heme between different cell compartments (Muller Eberhard and Nikkilä, 1989; Liem et al., 1994; Yuan et al., 2016). Thus, the specific functional roles of intracellular candidate HBPs in mammalians such as glutathione-S-transferases, heme-binding protein/ liver fatty-acid binding protein, heme-binding protein 23/peroxiredoxin 1, p22 heme binding protein and glyceraldehyde-3-phosphate dehydrogenase (Harvey and Beutler, 1982; Vincent and Muller Eberhard, 1985; Iwahara et al., 1995; Taketani et al., 1998; Chakravarti et al., 2010) need to be investigated in more detail (Table 1).

Enzymatic degradation of intracellular heme is primarily mediated via the HO system independent of P450s (Tenhunen et al., 1968; Maines and Kappas, 1974; Maines, 1997). The HO reaction has three major products: the signaling gas carbon monoxide (CO), iron and biliverdin, which is subsequently converted into bilirubin by biliverdin reductase (Kutty and Maines, 1981; Figure 2). Two genetically distinct isoforms of HO are known. HO-2 represents the constitutive non-inducible isoform and is primarily expressed in brain and testes (Trakshel et al., 1986). By contrast, the inducible HO isoform, HO-1, is expressed in almost all cells and tissues, and is highly up-regulated by heme or other stress stimuli to provide protection against oxidative damage and apoptosis (Maines, 1997; Ryter et al., 2006). The cytoprotective functions of HO-1 are directly linked with that of the iron-sequestering protein ferritin, which is co-ordinately up-regulated with HO-1 and neutralizes the pro-oxidant effects of the HO product iron (Balla et al., 1992; Figure 2). Importantly, genetic deficiency in mouse models and/or genetic deficiency and functional defects of HO-1 in humans are associated with major pro-oxidant and pro-inflammatory pathologies and with disturbed iron metabolism (Poss and Tonegawa, 1997a,b; Yachie et al., 1999; Kapturczak et al., 2004; Greil et al., 2016). In contrast, deficiency of the HO-2 gene in mice does not cause major heme-dependent pathologies (Poss et al., 1995) suggesting that HO-1 might be the more critical HO isozyme for counter-acting heme toxicity (Wagener et al., 2003; Kumar and Bandyopadhyay, 2005; Gozzelino et al., 2010). Studies in a conditional HO-1 knockout mouse model with targeted deletion in myeloid cells revealed a critical role of HO-1 for the regulation of innate immunity (Tzima et al., 2009). It is also interesting to point out that patterns of mouse and human HO-1 gene expression are differentially regulated in a species-specific manner (Sikorski et al., 2004). For example, HO-1 gene expression is up-regulated by lipopolysaccharide in mouse macrophages, but down-regulated in human macrophages (Miyazaki et al., 2010; Dorresteijn et al., 2015). Notably, the proximal promoter region of the human HO-1 gene, but not that of the mouse, contains a GT-microsatellite polymorphism, which may be responsible for interspecies-specific regulatory differences (Yamada et al., 2000). Finally, higher inducibility of the human HO-1 gene by oxidative stress has been associated with protection against cardiovascular disorders (Exner et al., 2004; Pechlaner et al., 2015) and against acute chest syndrome in SCD (Bean et al., 2012). Comprehensive overviews on the multiple roles of the HO system in health and disease have previously been given (Maines, 1997; Immenschuh and Ramadori, 2000; Abraham and Kappas, 2005; Ryter et al., 2006).

Hx protects against heme toxicity not only in animal models of hemolytic disorders such as SCD and malaria (Ghosh et al., 2013; Belcher et al., 2014), but also in other diseases, which are not typically associated with hemolysis such as sepsis (Larsen et al., 2010), cardiac disease (Vinchi et al., 2013) and bromine-induced acute lung injury (Aggarwal et al., 2016) (Table 2). It is plausible that the salutary effects of Hx described in animal disease models are translatable into the clinic. Hx might also have a prophylactic potential in clinical risk constellations, in which heme triggers overt disease manifestation. For example, pretreatment with Hx before transfusion of senescent RBCs improved the outcome in different mouse models of trauma-induced hemorrhage via neutralization of heme (Stapley et al., 2015; Graw et al., 2016). The role of Hx administration as a preventive intervention is also supported by findings of others (Tolosano et al., 2010). Hence, it is conceivable that patients that might benefit from prophylactic Hx treatment are those with comorbidities (i.e., diabetes mellitus, hypertension and older age) and an increased risk for AKI, with a high likelihood for the need of RBC transfusion during major surgery. Notably, in Japan Hp has been approved for medical indications, in which renal protection is required such as in massive transfusion or thermal injury (Schaer et al., 2012). Due to the known protective effects of Hx and Hp in hemolysis it is reasonable to assume that combined treatment with these two proteins may provide a synergistic protection in clinical settings of severe hemolysis (Schaer et al., 2014; Deuel et al., 2015; Graw et al., 2016).

HO-1 has been shown to provide specific protection against heme toxicity in different animal models including mouse models of SCD (Belcher et al., 2006, 2010), malaria (Pamplona et al., 2007; Seixas et al., 2009) and rhabdomyolysis (Wei et al., 2011). In particular, the beneficial effects of HO-1 have been demonstrated in mouse models, in which HO-1 has been either genetically deleted or overexpressed (Table 3). However, when HO-1 is targeted for therapeutic purposes, a major challenge is that HO-1 appears to be only protective when up-regulated before onset of an experimental injury. The latter has been confirmed in rodent models of experimental pancreatitis and colitis suggesting that HO-1 might be a primary option for prophylactic interventions (Nakamichi et al., 2005; Paul et al., 2005). Comprehensive overviews on challenges with potential translational applications of HO-1 in the clinic have recently been given for renal diseases (Lever et al., 2016).