Background
Severe malaria (SM), including severe malarial anaemia (SMA) and cerebral malaria (CM), is associated with high fatality rates [
1,
2]. Although the pathogenesis of SM remains unclear, fatal disease is associated with increased parasite burdens, excessive inflammation, endothelial activation, and multi-organ dysfunction [
1‐
3]. During malaria infection there is haemolysis of both parasitized and non-parasitized erythrocytes causing the release of cell-free haemoglobin. Cell-free haemoglobin is readily oxidized to methaemoglobin which releases plasma haem. Plasma haem is toxic to vascular endothelium via a number of established mechanisms, including iron-mediated oxidative stress, increased complement activation [
4], induction of pro-inflammatory mediators [
5], exocytosis of Weibel-Palade bodies [
4,
6], increased leukocyte and platelet adhesion [
6], and cell death [
7]. Under normal physiological states, cell-free haemoglobin and haem are promptly bound by circulating haptoglobin and haemopexin, respectively, leading to their inactivation via various pathways, including by haem-oxygenase 1 [
8]. However, in conditions of intravascular haemolysis these protective pathways may be overwhelmed resulting in oxidative stress, reduced nitric oxide bioavailability, inflammation, endothelial activation, and platelet/fibrin microthrombi that together culminate in vascular dysfunction and multi-organ injury.
Malaria causes haemolysis, however the key components of the haem axis, specifically haemopexin and its role, have not been well studied in SM. This study tests the hypothesis that severe malarial syndromes are associated with higher plasma haem and lower levels of haemopexin and haptoglobin. Using a case–control study the association of the haem axis with severe and fatal malaria in African children was investigated. These investigations were extended to explore a potential mechanistic role for haemopexin using a pre-clinical animal model.
Discussion
SM is a complex, multi-organ syndrome that is poorly understood. This study provides evidence that increased generation of plasma haem, combined with lower levels of haemopexin that normally binds and clears haem, are associated with both SMA and CM in Ugandan children. Furthermore, decreased or depleted levels of haemopexin, and increased plasma haem, are associated with a worse outcome in a pre-clinical model of CM. These observations support the hypothesis that haem-induced pathology, exacerbated by decreased levels of haemopexin, may be a pathway of injury contributing to disease severity and death in malaria infections.
The observed median plasma levels of haemin in children with CM and SMA were ten-fold or higher than that observed in UM [
12], and equivalent to those reported in haemolytic disorders such as sickle cell disease [
4], where haem-induced vasculopathy is a frequent cause of organ dysfunction and mortality [
8]. Although there were no differences in cell-free haemoglobin levels between groups, given the 11–15-fold higher levels of plasma haemin observed in children with SMA and CM and that cell-free haemoglobin generates haem, it is likely that there were also differences in levels of plasma haemoglobin during the course of infection, but the timing of sample collection did not capture these events.
Previously it has been shown that plasma levels of haem are higher on day 6 post infection in the ECM-susceptible C57BL/6 mice compared to ECM-resistant BALB/c mice [
13]. Injections of haemin into resistant BALB/c mice infected with
P. berghei ANKA resulted in increased fatalities compared to controls (100 vs 0 %) [
13,
14], suggesting that haem may play an important role in mediating, or enhancing, disease severity. While a role for haem has previously been implicated in ECM, it had yet to be confirmed in informative human populations. Recently, Dalko et al. showed that adults with SM in India had elevated levels of plasma haem compared to those with mild malaria, which were both higher than levels measured in endemic controls [
15]. These findings were extended in this study to show an association between increased haem and disease severity in African children with SM compared to those with UM.
It is hypothesized that haem is mediating adverse outcomes during malaria infection by inducing endothelial dysfunction and microvascular injury. Haem is implicated, directly or indirectly, as a key mediator of endothelial dysfunction by a variety of pathways. Plasma haem is a major source of reactive oxygen species (ROS) and subsequent pro-oxidant stress on endothelium [
8]. Haem also induces activation of the endothelium by: inducing Weibel-Palade body exocytosis and the release of their contents, including the release of von Willebrand factor (vWF), p-selectin [
4,
6], and Ang-2, and increasing expression of endothelial adhesion markers that bind both inflammatory cells and parasitized erythrocytes [
6]. Haem is also a pro-inflammatory molecule that can stimulate complement activation [
4] and the release of pro-inflammatory cytokines from monocytes/macrophages [
16]. Haem-induced endothelial injury may be further exacerbated by decreases in bioavailable nitric oxide. Cell-free haemoglobin in addition to generating plasma haem, is a potent scavenger of nitric oxide, further amplifying endothelial dysfunction [
8,
17]. Many of these pathways induced by haem are also implicated in the pathogenesis of SM [
3,
17‐
23]. Further prospective studies are required to define which of the above pathways are induced by haem in the context of SM.
The toxic effects of cell-free haemoglobin and haem are normally mitigated by high affinity-binding to haptoglobin and haemopexin, respectively, followed by transport and degradation. Haem is degraded by haem oxygenase-1 resulting in the generation of biliverdin and carbon monoxide which have anti-inflammatory and endothelial stabilizing effects [
8]. Therefore, the excess generation of haemin observed in children with CM and SMA appears to be further complicated by a relative deficiency in circulating haemopexin. Previous studies have primarily focused on levels of haptoglobin [
12,
24‐
26] or haemopexin [
24‐
26] in UM. However, it was recently shown that adults with SM in India had decreased levels of haemopexin compared to those with mild malaria [
15]. Similarly, in this study it was observed that children with SMA had significantly lower levels of both haemopexin and haptoglobin, in agreement with a previous report [
25]. This study extends these previous findings and reports that African children with CM also have lower levels of both haemopexin and haptoglobin, suggesting a common pathway of endothelial injury in both CM and SMA, and in both adults and children with severe disease. Despite the markedly low levels of these protective proteins in children with CM or SMA, there were no observed differences in either haemopexin or haptoglobin levels between children who survived or died of the infection. These data suggest that these proteins are already depleted during severe infection. Whether these relative deficiencies are due to consumption or due to a decreased capacity to synthesize these proteins in response to haemolysis is unknown. Nonetheless, these low levels would be expected to further compromise clearance and degradation of haem and haemoglobin, and contribute to worse clinical outcomes, as was illustrated by the observed higher plasma levels of haemin and cell-free haemoglobin in children who subsequently died of malaria.
Due to the associations observed in children with CM or SMA, a potential mechanistic role for haemopexin was investigated using a pre-clinical model of ECM. As with all animal models, there are inherent limitations. However, there is evidence that this model shares several features, but not all, with human malaria [
27‐
32]. Pre-clinical models enable investigation of mechanistic questions that are difficult to safely or ethically address in human populations and can provide direct genetic evidence that the haem axis is causally involved in severe and fatal infection.
Similar to this patient population, mice that were more susceptible to ECM had significantly lower haemopexin levels compared to more resistant mice, despite comparable parasite burdens. Furthermore, mice with targeted deletion of the hpx gene, and no measurable circulating haemopexin, were more susceptible to ECM than their wild-type littermates suggesting that haemopexin is playing a mechanistic role during malaria infection. The increased susceptibility in the Hpx KO animals was associated with increased levels of plasma haemin prior to the onset of ECM, further implicating a role for plasma haem.
A dose response to ECM was predicted due to the gene dosing effect observed in these mice; however, both the haemopexin-null and heterozygous mice responded similarly to ECM. The haemopexin-null mice appear to compensate for their lack of haemopexin by increasing the production of haptoglobin. The haemopexin-null mice have higher baseline levels of haptoglobin compared to both the wild type and heterozygous mice. When infected with P. berghei ANKA, haemopexin-deficient mice further upregulate circulating haptoglobin to at least 1300 times the levels seen in either the wild type or heterozygous mice. This compensation in a protein upstream of haemopexin in the haemolytic pathway is hypothesized to mitigate the severity of the observed phenotype. The haemopexin heterozygous mice may more accurately reflect the impact of haemopexin deficiency in ECM due to their inability to compensate with increased levels of haptoglobin to the levels observed in haemopexin-null mice. It would be of interest to further explore the phenotype of a double haemopexin, haptoglobin knockout mouse in the ECM model.
Despite the marked increase in haptoglobin, haemopexin-null mice are still more susceptible than their wild-type counterparts. This is likely attributable to malaria-induced generation of plasma haemoglobin and haem that overwhelms both haptoglobin and/or haemopexin pathways, as demonstrated by significantly higher levels of plasma haem prior to the onset of ECM in the KO animals. Alternatively, haemopexin may provide additional protection independent of its ability to clear haem from the circulation. Haemopexin has been shown to exert haem-independent, anti-inflammatory effects by decreasing the release of pro-inflammatory cytokines from lipopolysaccharide (LPS) stimulated macrophages [
33], and decreased differentiation of TH17 cells in a model of experimental autoimmune encephalitis [
34].
Collectively, the above observations suggest that malaria-induced haem, and corresponding haemopexin deficiency may represent a component of disease pathogenesis in which haem amplifies complement activation, pro-coagulant activity and endothelial injury, culminating in microangiopathy, multi-organ dysfunction and adverse clinical outcomes. The recent observation that both cases of CM and SMA are associated with parasitized erythrocytes that bind to endothelial protein C receptor (EPCR) lends further support to this hypothesis [
35]. Binding to EPCR may disrupt the anti-inflammatory and endothelial barrier effects of activated protein C (APC), resulting in enhanced pro-coagulant activity, complement activation and endothelial injury. Of note, APC mediates its endothelial cytoprotective and barrier protective activities via the Ang-Tie2 pathway [
36], which is dysregulated in SM [
37]. Moreover, since parasitized erythrocyte sequestration focuses parasite burdens and erythrocyte haemolysis directly onto the microvasculature of vital organs, this would be expected to result in higher local haem concentrations in the cerebral microvascular bed, than those observed in the peripheral circulation.
Similar to atypical haemolytic uremic syndrome (aHUS) associated with infection, where haem-induced complement activation is a proposed common pathway of microvascular injury and organ failure, severe disease does not occur in all infected individuals, but rather only in those with genetic or acquired defects in complement regulation [
4]. This suggests that similar or related susceptibility determinants, including defects in haemopexin expression or haem metabolism, may contribute to the onset and outcome of severe malarial syndromes.
Authors’ contributions
KCK, HSW, REE, FR, TL, and SH conceived and designed the experiments. AD, CM and CCG completed the larger observational prospective study and provided clinical samples for analysis. FR, TL and REE measured plasma levels of the haem axis. REE and SH performed the ECM experiments. REE performed the statistical analysis. HSW, KCK and RR provided reagents and materials for the experiments. REE and KCK drafted the manuscript. All authors read and approved the final manuscript.