Malaria, anemia, and invasive bacterial disease: A neutrophil problem?

Abstract Invasive bacterial disease is well described in immunocompromised hosts, including those with malaria infection. One bacterial infection frequently observed in children with Plasmodium falciparum infection is nontyphoidal salmonella (NTS) infection, in which a typically intestinal infection becomes systemic with serious, often fatal, consequences. In this review, we consider the role of malaria‐induced immunoregulatory responses in tipping the balance from tissue homeostasis during malaria infection to risk of invasive NTS. Also, neutrophils are crucial in the clearance of NTS but their ability to mount an oxidative burst and kill intracellular Salmonella is severely compromised during, and for some time after, an acute malaria infection. Here, we summarize the evidence linking malaria and invasive NTS infections; describe the role of neutrophils in clearing NTS infections; review evidence for neutrophil dysfunction in malaria infections; and explore roles of heme oxygenase‐1, IL‐10, and complement in mediating this dysfunction. Finally, given the epidemiological evidence that low density, subclinical malaria infections pose a risk for invasive NTS infections, we consider whether the high prevalence of such infections might underlie the very high incidence of invasive bacterial disease across much of sub‐Saharan Africa.

lack of access to effective and affordable antibiotics may lead to even higher morbidity and mortality in low-income settings.
NTS thrives in the intestinal environment where, in otherwise healthy hosts, localized gastroenteritis allows NTS to outcompete the microbiota, causing diarrhea and promoting transmission. 15 However, the infection can "escape" the gut and invade other tissues, eventually becoming systemic, particularly when the host is immunocompromised. One well-documented risk factor for invasive NTS is Plasmodium falciparum malaria. 16,17 Plasmodium, the causative agent of malaria, is transmitted to humans through the bite of the female Anopheles mosquito causing a range of clinical manifestations including anemia, metabolic acidosis, and end-organ failure. 18 In The Gambia, the incidence of invasive NTS infection mirrors that of malaria, peaking during the annual rainy season, and in one study, 43% of children with Salmonella bacteremia had concurrent P. falciparum infections. 19 In Tanzania, invasive NTS in young children is highly associated with recent malaria infection, with 78% of NTS cases having recently received antimalarial medication and 82% of cases being anemic. 20 Intriguingly, recent (past) malaria infection is a higher risk factor for NTS bacteremia than is acute (current) infection. 21 Therefore, although children with severe acute malaria have been noted to be at high risk of developing invasive NTS, 22 a picture is emerging in which even low-density or recently cleared malaria infections are a significant contributor to invasive NTS. Finally, evidence that carriage of sickle cell trait (that protects from malarial anemia) reduces the risk of contracting invasive NTS 23 and that intensive efforts in the last 15 years to reduce the prevalence and incidence of malaria across Africa have been accompanied by marked falls in the incidence of invasive bacteremia, and especially invasive NTS 24,25 serves to reinforce the clinical observations linking these two diseases and suggests a related underlying pathophysiology.

INTESTINAL AND INVASIVE NTS INFECTIONS
Salmonella can infect a broad host range (e.g., pigs, cattle, chickens, and humans) causing varying levels of damage, from enteric fever to severe gastroenteritis to asymptomatic carriage, depending on the particular serovar, typically defined by expression of LPS, flagellar, and capsular Vi antigens. 26 With over 2500 known serovars, 27 sterile immunity through natural infection or vaccination remains elusive. 28,29 The human-restricted typhoidal serovars (Salmonella typhi and Salmonella paratyphi) are associated with systemic infection and carriage in the gallbladder 30 but, intriguingly, these are not the serovars that are associated with malaria infections. Rather, malaria is associated with invasive disease caused by nontyphoidal serovars that can infect a broad range of different host species and are normally restricted to the intestine. 17 Invasion of NTS through the intestinal mucosa can occur via their uptake by dendritic cells extruding dendrites between enterocytes into the intestinal lumen (paracellular uptake), via direct invasion of enterocytes or by passage through M cells of the Peyers Patches 31 (Fig. 1), and this is dependent on a degree of inflammation. 32,33 Bacterial invasion triggers the IL-23/IL-18 inflammatory axis leading to T cell activation and production of inflammatory cytokines (including IFN-, IL-17, and IL-22) and chemokines (CXCL1 and Mip2), [34][35][36] eventually resulting in edema and infiltration of monocytes and neutrophils into the lamina propria, 37 which are hallmarks of NTS pathology.
Although phagocytes, such as neutrophils and macrophages, are efficient in their uptake of NTS, the bacteria can disable the antibacterial machinery of macrophages to create Salmonella containing vacuoles (SCVs) within which they can persist and replicate. 38 Proteins encoded within Salmonella pathogenicity island-2 block lysosomal fusion allowing evasion of ROS-mediated killing. 39 Ultimately, clearance of bacteria from phagocytes is mediated by IFN-, which induces breakdown of the SCV, 40 releasing bacteria into the cytosol.
Bacterial products now present in the cytosol can induce pyroptosis, a form of cell death involving both canonical and noncanonical inflammasome signaling with caspase-1 and NLRP3 or caspase-11 (in mice) and caspase-4 and caspase-5 (in humans), which lead to activation of IL-1 and IL-18. [41][42][43] Bacteria released from disintegrating macrophages are cleared by neutrophils, a process that has been termed "phagocyte roulette." 44 Neutrophils, also called polymorphonuclear cells (PMNs), are terminally differentiated leukocytes with distinctive lobulated nuclei and contain antimicrobial cytoplasmic granules. PMNs are the most abundant white blood cell, with 1 × 10 11 new cells emerging from the bone marrow daily. They are typically thought to have a very short lifespan in blood (∼7-24 hours 45 ), although infection may delay apoptosis and increase lifespan. [46][47][48] During infection and inflammation, PMNs are quickly mobilized to sites of injury. In the blood vessel, activated PMNs adhere to endothelium, extravasate and migrate along chemokine gradients to infectious foci.
PMNs are professional phagocytes, which use receptormediated phagocytosis to internalize pathogens and debris into phagolysosomes. 49 Intracytoplasmic granules containing cathepsins, elastases, and myeloperoxidases fuse with the phagolysosome to digest internalized pathogens, in a process known as degranulation; leakage of granules or their contents into the extracellular milieu can be a significant cause of tissue damage during infection. 49,50 Release of reactive oxygen species (ROS), produced via an NADPH oxidase-dependent process, into the phagolysosome is an additional, very important, bactericidal mechanism. PMNs can also kill extracellular pathogens by degranulation, secretion of ROS, or the release of neutrophil extracellular traps (NETs). NETs consist of externalized decondensed chromatin decorated with granular proteins and histones to prevent the dissemination of pathogens. 51,52 Serine proteases and histones provide antimicrobial activity against trapped pathogens. NETs also permit subsequent phagocytosis by proximate phagocytes (Fig. 2).
Although macrophages can be permissive to NTS growth, inflammatory monocytes and neutrophils are efficient in killing Salmonella through oxidative stress. 53 In the intestine, both mucosa-associated and luminal neutrophils engulf Salmonella. 54 However, infiltrating PMNs also promote intestinal inflammation, 55 NTS is a fecal-oral pathogen, which thrives in the inflamed intestine. Tissue invasion in the distal intestine can be via direct invasion, uptake by M cells, or through paracellular spaces. Uptake by dendritic cells can initiate inflammation through the IL-23/IL-17 axis. Th17 cells promote neutrophil influx via the induction of neutrophil chemokines. NTS is able to persist within the Salmonellacontaining vacuole of macrophages, whereas neutrophils are efficient at NTS clearance. Systemic dissemination to draining lymph nodes is through CD18 + phagocytes. During experimental malaria, NTS colonization resistance is lowered, although it is unclear if there is an increase in tissue invasion. Regardless, inflammation (with reduced PMN influx) is reduced due to increased IL-10 concentrations. However, the role of IL-10, and potentially HO-1, on intestinal neutrophil function and role for increased systemic dissemination are unclear into tetrathionate and the microbial fermentation product 1,2propanediol, allowing NTS to outgrow the competing microbiota. 32,33 PMNs also release calprotectin into the intestinal milieu where it sequesters zinc, further restricting the growth of the intestinal microbiota. 56 Therefore, although PMNs can limit bacterial growth and prevent overwhelming infection, NTS can exploit neutrophilmediated inflammation in the intestine to ensure its survival and eventual transmission. 53,57

PATHOPHYSIOLOGY OF NTS-MALARIA COINFECTIONS
Clinically, NTS infections in patients in Africa are not associated with overt diarrheal disease 58 suggesting that underlying coinfection may ameliorate the intestinal inflammation typically associated with NTS. Evidence from coinfection models supports this idea ( PMN F I G U R E 2 Neutrophil function. PMN extravasation is initiated by inflammatory mediators, which stimulate the upregulation of adhesion molecules, such as selectins, on endothelial cells allowing tethering and rolling of PMN. Chemokine gradients cause PMNs to "crawl" toward the site of infection until an endothelial junction is reached where diapedesis can occur. At the site of infection, neutrophils have numerous functions including phagocytosis, degranulation, and NETosis. Phagocytosis involves the ingestion of a pathogen into a phagolysosome containing granule proteins and ROS, to disable and digest internalized pathogens. Degranulation can also occur extracellularly where granular proteins, such as MPO and elastase, and ROS can be released from PMNs. A novel function of PMNs is NETosis, which is the release of NETs. These decondensed chromatin fibers decorated with histones and granular proteins trap and kill pathogens to prevent further dissemination. It is known that during malaria infection, neutrophils maintain the ability to phagocytose NTS; however, PMNs display impaired ROS production. Mechanisms still to be investigated include: maturation, extravasation, migration, degranulation, and NETosis and macaques coinfected with malaria and NTS compared to animals infected with NTS alone and this is associated with reduced neutrophil influx and lower levels of IFN-and IL-17 59 (Fig. 1). In mice, this reduction in intestinal inflammation is mediated by IL-10, 59  however, complement-mediated opsonization of bystander RBC has been noted. 66,67 Regardless of the mechanism, hemolysis results in the release of hemoglobin, or its breakdown product heme, into the plasma (Fig. 3). Free heme is prooxidant and highly cytotoxic, contributing to endothelial injury and hypertension. 68 Anemia, HO-1, and IL-10. During malaria infection, anemia can occur through direct lysis of parasitized RBCs (hemolysis) or through eryptosis of uninfected RBCswhich are taken up by phagocytes through the phosphatidylserine receptor (PSR). Hemolysis results in release of hemoglobin (Hb), which can be further degraded to free heme. Serum scavenger proteins haptoglobin (Hpt) and hemopexin (Hpx) bind Hb and heme before uptake by phagocytes through CD163 and CD91, respectively. Intracellular heme is then degraded to iron, biliverdin, and carbon monoxide via the enzyme HO-2 or the inducible isoform, HO-1. HO-1 can also be induced by IL-10 receptor signaling. The impact of IL-10 and HO-1 during malarial anemia and hemolysis on phagocyte function remains poorly defined The first indication that hemolysis has a negative impact on neutrophil function, and thus that neutrophil dysfunction might underlie increased risk of invasive bacterial disease in people with malaria or other hemolytic diseases, came from mouse coinfection studies where circulating neutrophils from malaria-infected and phenylhydrazinetreated mice were shown to efficiently phagocytose S. Typhimurium but were unable to kill; viable bacteria persisted and, indeed, replicated inside neutrophils, which were severely deficient in ROS production. 63 Of note, in vitro, heme pretreatment reduced phagocytosis of E.
coli by human and murine neutrophils 78 and high circulating heme during P. falciparum infection can reduce in vitro phagocytosis of Salmonella by neutrophils. 72 The observation that treatment with a synthetic heme polymer, hemin, induced similar neutrophil defects and that these defects can be reversed by competitive inhibition of HO-1, then provided a link among malaria, hemolysis, HO-1, and neutrophil dysfunction. 63 ROS-defective neutrophils have also been observed in children with acute malaria; importantly, these defects persist for up to 8 weeks after treatment, 72 perhaps explaining why children with recent (past) malaria infection remain susceptible to invasive NTS. 21 Plasmodium infection also reduces neutrophil mobilization into infected tissues including blood 63 and, as discussed above, intestine 59 and liver 64 ; infiltration of inflammatory monocytes into the liver is also impaired. 79 Although this may be due in part to the anti-inflammatory effects of IL-10 (as described below), there are also cell-intrinsic effects within developing neutrophils. Neutrophil precursors in the bone marrow (i.e., granulocyte macrophage progenitor cells) of malaria-infected mice express HO-1 and have unusual surface phenotypes (F4/80 and Gr-1 expression). 63 It has been shown that HO-1 reduces neutrophil influx into the inflamed lung 71 suggesting a causal relationship between HO-1 and reduced neutrophil migration but much more work is needed to fully characterize neutrophil maturation and function during malaria infection and to determine the extent to which the altered phenotype is mediated by the heme/HO-1 pathway.

THE ROLE OF IL-10 IN MALARIA NTS COINFECTION
IL-10 is a potent anti-inflammatory cytokine and an important regulator of inflammation-induced pathology, 80 it is therefore no surprise that systemic IL-10 concentrations are elevated in highly inflammatory diseases such as sepsis 81,82 and malaria. 83,84 More surprisingly, however, circulating IL-10 is also elevated during mild/uncomplicated 83,85 and asymptomatic/subclinical malaria infections 73,86 ; indeed these cases may represent successful balancing of inflammation-mediated parasite control and effective regulation of inflammation by IL-10. In malaria, IL-10 can come from both innate and adaptive immune cells, including Th1-derived regulatory T cells that coproduce IFN-and IL-10, [87][88][89] and plays an essential role in both adaptive humoral immunity (promoting differentiation of T-bet + germinal center B cells) 90,91 and limiting tissue damage. [92][93][94] The anti-inflammatory properties of IL-10 include rendering phagocytes refractory to activation and/or directing macrophage polarization to a regulatory "M2" phenotype. 95,96 The impact of IL-10 on neutrophil function is well described, reducing recruitment and migration in response to anaphylatoxins and ultimately bacterial clearance. 97,98 Further, neutrophils themselves can be a source of IL-10, induced by regulatory T cells. 99 During malaria NTS coinfection in mice, LysM-expressing cells are a significant source of IL-10 and its ablation reduces NTS bacteremia. 64 IL-10 suppresses neutrophil function through the activation of STAT3 and suppressor of cytokine signaling 3, 100,101 leading to the down-regulation of IFN regulatory factor and NF-B family transcription factor. 102 In addition, HO-1 can be directly induced by IL-10 103 (Fig. 3). One hypothesis, therefore, is that malaria-induced neutrophil dysfunction results from hemolysisand IL-10-driven HO-1 induction. In support of this hypothesis, in a recent study of persistent malaria infections, we have found that heme drives inflammation, that parasite density and inflammation then drives IL-10 production, and that heme and IL-10 both then induce HO-1. 73

COMPLEMENT DEPLETION DURING MALARIA
Complement proteins play an essential role in orchestrating inflammation and pathogen clearance, 104,105 including during Salmonella infection. 106,107 In brief, the classical pathway is activated by antibody (IgM and IgG) and the alternative pathway is activated by spontaneous hydrolysis of C3. 108 C3 becomes membrane bound and cleaves C5, creating soluble C5a and membrane-bound C5b. 109 The complement anaphylatoxin C5a is a potent chemoattractant for neutrophils while also inducing expression of cell adhesion molecules on endothelial cells (such as P-selectin) that permit neutrophil translocation into tissue. 110 C5a also enhances neutrophil resistance to apoptosis, 111

CONCLUDING REMARKS
We have provided a rational for considering that hemolysis and inflam- (1) a reduction in C3 available for deposition on extracellular NTS leading to opsonization and migration and/or (2) excessive C5a reducing CD88 (C5aR) on neutrophils leading to reduced ROS and cytokine production. Additional work is needed to clarify the impact of complement activation and subsequent depletion during malaria on risk to NTS bacteremia susceptibility to bacterial coinfection. It is likely that these pathways synergize to increase risk of invasive NTS: in malaria-infected mice, both exogenous IL-10 and anemia were required for increased bacteremia 64 ; HO-1 is induced both by heme and by IL-10; and the carbon monoxide generated from heme catabolism can induce both HO-1 and IL-10. 121,122 Ultimately, it is important to define how these factors may converge to alter neutrophil biology. Through this, we may begin to understand if targeted treatment, or even prophylactic antimalarial treatment, can improve neutrophil function and so reduce the burden of invasive bacterial disease in malaria endemic populations. Given the importance of neutrophils in clearance of NTS, work is now needed to better describe the impact of malaria on this innate immune cell.
The risk of invasive NTS during malaria in sub-Saharan Africa is well defined for acute malaria infection. However, the majority of malaria infections in the world are asymptomatic, with chronic, low-density infections. 123 As recent and low-density malaria infections are a risk factor for NTS bacteremia, 21 it is also important to understand if hemolysis, and resulting induction of HO-1 and IL-10, seen during these "asymptomatic" infections 73 reaches the threshold needed for neutrophil dysfunction. In other words, in addition to defining the pathways leading to neutrophil dysfunction, we also need to identify the point at which the balance tips from these being host protective to increasing the risk to invasive NTS. Importantly, these pathways may contribute to severe bacterial disease even in the absence of malaria infections: in sepsis patients, we have observed that raised concentrations of heme, HO-1, and IL-10 are positively correlated with disease severity and mortality. 82 Also, while "invasive" NTS is seen in immunocompromised hosts (such as those with malaria infection), it remains unclear if this is due to increased intestinal invasion, increased dissemination from draining lymph nodes, failure to control systemic bacterial replication, or a combination of any of these. Moreover, the extent to which this is primarily a neutrophil defect requires further exploration.

DISCLOSURES
The authors declare no conflicts of interest.