Any theory about the development of the CM syndrome must involve either the transfer of active mediators from the blood into the brain tissue, or a parenchymal response to signals coming from inside blood vessels, making the blood-brain barrier a crucial interface in this syndrome (Adams, 2002).
The blood brain barrier (BBB) is highly specialised structural and functional interface between the intravascular space and the CNS. It is composed of EC, basement membrane, pericytes, and astrocyte foot processes, and is set up to tightly limit and polarise molecular and cellular trafficking into the brain (Medana and Turner, 2006). The high mortality associated with CM is thought to be related to the ability of the infection to disturb the integrity of the BBB. Histological, immunohistochemical, and ultrastructural studies of post-mortem brain tissue from fatal cases of human CM and in the mouse model have revealed structural changes at the BBB (Adams et al., 2002). Examples of these include evidence of EC activation, upregulation of ICAM-1 and E-selectin, and other constitutively expressed adhesion molecules, and induction of soluble mediators from other CNS cells (Turner et al., 1994; Medana and Turner, 2006).
The process by which the parasite is thought to cross the BBB is through P. falciparum erythrocyte membrane protein-1 (PfEMP-1). This protein mediates pRBC adhesion to ICAM-1, and this increases junctional permeability of the BBB and suppresses dendritic cell and macrophage activation (Adams et al., 2002; Schofield and Grau, 2005; Medana and Turner, 2006; Faille et al., 2009b). Another theory is that soluble products, such as cytokines and parasite toxins, could cause the release of cytokines and other mediators by host leucocytes. For example, TNF and LT induce changes in the BBB structure (Hunt and Grau, 2003). Additionally, local but adhesion-independent changes at the BBB such as the effects of parasite toxins, decreased blood flow, resulting in hypoxia, may activate or damage cerebral EC and perivascular cells.
In the murine model, there are two distinct phases of BBB changes: firstly there is a mild localised breakdown three days after infection, and then, seven days after infection, more extensive disruption occurs, leading to plasma leakage into the perivascular space, activation and alteration of the distribution of astrocytes and glial cells, as well as the changes previously mentioned such as upregulation of endothelial adhesion molecules, accumulation of sequestered cells, increase in BBB permeability, cytokine stimulation leading to activated inflammatory processes disturbing neuronal function and leading ultimately, to death (Medana et al., 1996; Medana et al., 2000).
Oxygen is a requirement for normal function of mammalian cells, in order to maintain homeostasis, and providing the basis for aerobic metabolism. Transcriptional changes can occur quickly, to reduce hypoxia-associated tissue damage, for example the rapid upregulation of the transcription factor hypoxia-inducible factor (HIF)-1 in response to hypoxic conditions (Sharp and Bernaudin, 2004). Impaired cerebral microcirculation has previously been shown to be caused by vascular occlusion due to sequestered pRBC, platelets, and leucocytes, among other factors. Cerebral hypoperfusion is one consequence of this microcirculatory blockage, and failure to adapt to hypoxia resulting in irreversible tissue damage and a clear association with poor clinical outcomes for patients (Dondorp et al., 2008; Beare et al., 2009). In the mouse model, increased intracranial pressure and correspondingly decreased cerebral blood flow have been identified (Penet et al., 2005), leading to cerebral ischaemia (Rae et al., 2004) and potentially hypoxia (Polder et al., 1991; Cabrales et al., 2013). Reversing this therapeutically, for example using erythropoietin, which is a hypoxia-responsive hormone, decreases cerebral disease and improves survival (Kaiser et al., 2006; Wiese et al., 2008). Hempel et al. demonstrated the presence of multifocal areas of cerebral hypoxia in eCM, detecting widespread low-grade intercellular and intracellular hypoxia, as well as specifically in neuronal and glial foci (Hempel et al., 2011).
Astrocytes are essential components of the BBB, and function to modulate synaptic transmission and the ionic composition of the brain, and to control metabolic processes and microvascular behaviour. Astrocytes can become dysfunctional in response to pRBC sequestration and parasite molecules, as well as hypoxic conditions and increased cytokine stimulation, resulting in apoptosis of these and microglial cells (Deininger et al., 2002). In fact, axonal and astrocytic injury markers were found increased in CSF samples from CM patients (Medana et al., 2007). Oxidative stress and neuronal injury has also been demonstrated in the vicinity of vessel haemorrhaging (Medana et al., 2001). During CM, alterations in BBB function incite changes in the distribution and properties of astrocytes and microglia, both key non-neuronal elements in the brain parenchyma (Medana et al., 1997a). Microglia have been shown to become activated within 48-72 hours post PbA infection and, later in the infection course, take on an amoeboid appearance, which is representative of an immunologically activated state (Medana et al., 1997a), and subsequently express TNF (Medana et al., 1997b).
The best way to study the various elements of CM pathogenesis and their interactions with each other, is to examine them in real time during the infection, using techniques such as depletion studies, directed therapies, adoptive transfer, or microscopy or flow cytometry-based approaches.
Depletion studies, and adoptive transfer experiments have been widely used in order to ascertain whether specific cell types, vesicles, or soluble products contribute to disease progression in CM, and their role during infection. Interestingly, specifically depleting certain cytokines or cells during PbA infection prevents CM, however this success is dependent on the timing of depletion and, to some degree, the mouse strain used, as detailed in section 2.4 (Chang et al., 2001; van der Heyde et al., 2005; Schumak et al., 2015). Directed therapies, similarly to depletion studies, can be used to target individual cell types or groups. Though most treatment options for CM are directed at the parasite, immune-targeting therapies provide a new avenue to address the syndrome (Zumla et al., 2016). Adoptive transfer has been utilised in the opposite manner, in order to see whether introducing particular cell types or cytokines can induce disease in healthy mice, or worsen disease progression in infected mice. In this technique, cells can be harvested from donor mice and sorted out of a heterogeneous population of cells based on specific light scattering and fluorescent characteristics of each cell using fluorescence-activated cell sorting (FACS). These sorted cells, or other molecules can then be reinjected into donor mice, and the effects monitored (El-Assaad et al., 2014).
Fluorescence microscopy-based approaches have been used to study circulating and tissue-resident cells in the brain and other tissues for many years; however, recently, new and improved techniques of optical imaging, such as 2-photon microscopy, and clearing techniques have allowed tissue sections of increasing size to be examined, up to whole tissues or whole organisms, through ex vivo or intravital imaging (Pai et al., 2014; Susaki et al., 2015). As the technology continues to improve, likewise the speed and resolution at which this imaging can take place is rapidly improving.
Finally, multi-colour flow cytometry allows us to specifically label many different cell types and identify changes in number and surface markers in response to infection, with increased sensitivity and complexity compared with confocal microscopy (Basiji et al., 2007). The number of markers able to be tested in one panel is continually increasing, with 18 markers being the current demonstrated maximum (Ornatsky et al., 2010). More recently, Cytof technology (mass cytometry) has emerged, where antibodies are labelled with heavy metal ion tags instead of fluorochromes, enabling many more markers to be assessed simultaneously in a single sample, without significant spectral overlap between channels. Up to 30 unique markers have been assessed at one time, with up to 60 distinguishable labels theoretically possible (Ornatsky et al., 2010). However, the practical flow rate is considerably lower than in traditional flow cytometry (Fluidigm, 2017).
Studying CM pathogenesis in humans is only recently beginning to be implemented, through studies utilising controlled human malaria infections (Spring et al., 2014). Post-mortem studies are another alternative, however limited information is provided by such studies as only the endpoint of the disease can plainly be analysed. Therefore in vivo approaches, such as use of animal models and in vitro techniques, provide a useful tool in further understanding the disease. The main challenge facing researchers using animal models is to replicate the features of the disease as accurately as possible. Models exist in various primates, and rodents such as hamsters, rats, and mice.
Primates are naturally affected by malaria but not much is known about the outcome of the disease in this situation. Rhesus monkeys are typically infected with P. knowlesi, P. coatneyi, or P. fragile for experimental study. Monkeys develop acute symptoms a week p.i. show cerebral vascular congestion, pRBC sequestration in the brain, rosette formation, and increased levels of ICAM-1, TNF, and IFN- similar to human CM (de Souza and Riley, 2002). Therefore, as the pathological changes so closely resemble human CM, and as malaria infections occur naturally in primates, this model is useful for comparison and study (Lou et al., 2001). However, some limitations exist with this model: determining the time of CM onset is difficult, and the incidence of this syndrome is low and somewhat unpredictable in the monkeys. In addition, there are ethical considerations, significant financial expenditure, and lack of genetically modified animals (Combes et al., 2005a).
Hamsters and rats have also been used as experimental models, but infrequently. The majority of rodent experiments involve mice infected with PbA, which reproduces over 25 pathophysiological, biochemical, clinical, histopathological, and immunological features of human CM (Lou et al., 2001; Hunt and Grau, 2003; Combes et al., 2005a; Hunt et al., 2010; Riley et al., 2010; Craig et al., 2012). For example, the same behavioural changes, histopathological features and expression of molecules in the brain and retina, and changes to blood-brain barrier function can be seen in paediatric, adult, and murine CM (Hunt and Grau, 2003). The rodent and human parasites also share defined malaria antigens. Moreover, mice are quicker and more simply bred, and many different strains and genetic modifications are known to be susceptible or not to CM. A higher incidence of the disease occurs in mice, and the disease tends to follow a more easily predictable infection course (de Souza and Riley, 2002). This model, compared with that of monkeys, is more easily managed and widely available.
Known CM susceptible mice include CBA, C57BL/6, and ABCA-1+/+ DBA/1 mice, and these mice reproducibly die between days 7 and 10 following PbA infection (clinical scores of 3-4) (Lou et al., 2001). Known resistant mice include BALB/C and ABCA-1-/- DBA/1 mice, and these mice reproducibly die during the third week after infection, but do not develop CM, but instead are affected by severe anaemia and hyperparasitaemia (clinical score 2) (Lou et al., 2001).
Despite the large amount of insight the murine models provide into the disease, some disadvantages do exist. There are obvious anatomical and physiological differences between mice and humans, and this may impact experimental findings (Lou et al., 2001). Obviously, primates are a closer approximation of this, but, as detailed above, there are a number of other issues with that model. In the murine model, pRBC cytoadherence does not occur for all parasites, the syndrome cannot be reversed, and some parasitological differences exist between murine and human parasites (de Souza and Riley, 2002). In addition, P. falciparum does not infect mice, limiting the comparisons with humans (Combes et al., 2005a). However, even considering these disadvantages, the murine model remains a valuable tool in discovering more insight into the pathogenesis of CM.
Microparticles, now called microvesicles (MV), are submicron elements that originate from a loss of phospholipid asymmetry associated with the remodelling of the cytoskeleton and the increase in cytosolic calcium (Zwaal and Schroit, 1997). They are cellular-derived vesicles, and range from 0.2 – 1 m in size (Combes et al., 2010), and are also described in the literature as microparticles, ectosomes, exovesicles and shed vesicles (Doeuvre et al., 2009; Burger et al., 2013). All cell types have the capacity to vesiculate (produce MV) under normal physiological conditions, upon cellular activation and during apoptosis (Coltel et al., 2006). MV released from platelets, EC, RBC, and monocytes have been studied extensively in comparison to other cell types. Wolf first observed them in 1967, where he referred to them as “platelet-dust.” They were described as fragments derived from platelets in human plasma, and were thought to be a residue or by-product of platelet activation, hence the name platelet-dust (Wolf, 1967).
Much of the structure of microvesicles has been elucidated from electron microscopic study. This reveals that MV exist in varying sizes and densities (Combes et al., 1999). MV membranes consist of mainly lipids and proteins, but their composition depends on their cell of origin, and the cellular processes that trigger their formation. MV are known to be compromised of cytoplasmic molecules, cytoskeletal components, enzymes, cofactors, and various surface antigens unique to the cells from which they originate and the mechanisms by which they are formed (Coltel et al., 2006).
Each MV has a bilayer membrane asymmetrically distributed with positively and negatively charged phospholipids. Phosphatidylcholine (PC) and sphingomyelin (SM) – both positively charged phospholipids – are located on the outer leaflet of the membrane, whereas phosphatidylserine (PS) and phosphatidyl-ethanolamine (PE) – negatively charged phospholipids – are located on the inner leaflet of the membrane (Hugel et al., 2005). The membranes of MV contain lipids and proteins carrying markers from their cell of origin, and these lipids and proteins are dependent on the cellular processes responsible for their formation (Zwaal and Schroit, 1997; Piccin et al., 2007).
Analysis of the proteomic, lipidomic, and nucleic acid content of MV provides a strong basis for future studies to understand MV biology and pathophysiology (Weerheim et al., 2002; Yoon et al., 2014; Tiberti et al., 2016). In disease states, MV derived from injured organs likely contain valuable markers for determining the site, type, and extent of disease pathology. However, the basic protein characteristics of plasma MV are still being explored. Tissue factor (TF) activity is known to be present in MV purified from healthy individuals, confirming the role of MV in initiating blood coagulation (Jin et al., 2005). MV proteome analysis is carried out using gel electrophoresis, liquid chromatography, and mass spectrometry, and these results can aid the identification of different proteins found in MV. Jin et al. found that MV displayed distinct protein features compared to whole plasma samples (Jin et al., 2005). Garcia et al. found that MV displayed many proteins intrinsic to and well characterised on platelets, for example P-selectin (Garcia et al., 2005).
Significant proteomic variances have been observed between MV of different origins. Plasma-derived MV express 21 additional proteins involved in cell processes such as apoptosis, immune responses, and coagulation, compared with platelet MV (Smalley et al., 2007). Additionally, MV produced from drug-resistant cancer cells contained P-glycoprotein, which is normally overexpressed on the surface of these cells, and these MV go on to affect drug-sensitive cancer cells by transferring this protein (Bebawy et al., 2009; Jaiswal et al., 2012). Here, it was shown that multi-drug-resistant (MDR) MV incorporate nucleic acids, that MV change recipient cells’ transcriptional environment to mimic donor MDR phenotype, and distinct pathways exist among cancers of different origin that may be dependent on donor cells’ ABCA1 overexpression (Jaiswal et al., 2012).
MV are produced via the process of vesiculation, which occurs when numerous enzymes, proteins, and cellular components coordinate to ultimately disrupt the plasma membrane and result in blebbing (Freyssinet, 2003).
The translipid bilayer distribution of the plasma membrane is under the control of three enzymes. The first of these is an inward-directed pump or a flippase, specific for PS and PE, called aminophospholipid translocase. The second is an outward-directed pump referred to as “floppase”; and the third is a lipid scramblase, responsible for promoting random movement of phospholipids in both directions across the bilayer (Freyssinet, 2003; Hugel et al., 2005). Translocase and floppase are adenosine 5’-triphosphate (ATP)-dependent (Piccin et al., 2007). The increase in cytosolic calcium, as described above, occurs during cell activation and apoptosis, and stimulates the random movement of phospholipids across cell membranes though the action of scramblase, a membrane enzyme, while inhibiting the translocase enzyme (Zwaal and Schroit, 1997; Martinez et al., 2005; Piccin et al., 2007). Other enzymes involved in this process are gelsolin, which contributes to actin reorganisation, and calpain, which cleaves cytoskeletal filaments, facilitating MV shedding (Piccin et al., 2007).
When cells are subjected to procoagulant, proinflammatory, or apoptogenic stimulation, a spontaneous collapse of their membrane asymmetry typically occurs. In the resting state, as calcium levels are low, scramblase is inactive, and ATP-dependent translocase and floppase are responsible for maintaining phospholipid asymmetry (PS and PE still on the inside of the membrane). As calcium is released by the endoplasmic reticulum, scramblase is activated, and translocase and floppase are inactivated, leading to the beginning of phospholipid asymmetry being compromised. Following cellular activation, cytoskeletal disruption occurs, structural proteins are distorted, phospholipids are reorganized, and, through the “flip-flop” mechanism, PS migrates from the inner to the outer leaflet. Finally, vesiculation is complete, with the generation of MV (Piccin et al., 2007; Wassmer et al., 2011a).
All through this process, calcium acts as an agonist triggering cell activation. Calcium levels rise in response to cell activation, and this in turn stimulates the activation of calpain, and gelsolin. Calpain cleaves cytoskeletal filaments, facilitating MV shedding, and activating apoptosis. Gelsolin disturbs actin filaments in platelets, also contributing to the reorganisation of the membrane, and thus, MV formation (Zwaal and Schroit, 1997; Piccin et al., 2007).
The ATP-binding cassette (ABC) transporters are members of one of the largest families of proteins. These transporters require ATP to overcome the substrate concentration gradient to transport substrate through the membrane. The ATP-binding cassette transporter A-1 (ABCA-1) is a prototype of ABC family subclass with a function to transport lipids and other metabolites across plasma membrane (Hamon et al., 2000). ABCA-1 has been repeatedly implicated in processes that are likely to be affected by dynamic distribution of lipid species across the membrane bilayer. ABCA-1 functions at the plasma membrane as a floppase of PS with the net result of increasing the amount of this lipid in the outer leaflet, hence suggesting that ABCA-1 activation promotes the release of microvesicles from the plasma membrane, as demonstrated in mice deficient in the ABCA1 gene (Hamon et al., 2000; Combes et al., 2005b).
It is known that, thus far, all MV express PS on their surface, independent of their cell of origin. Therefore, this provides a target for MV detection. One such detection method is annexin V labelling. Annexin V is a selective ligand for negatively charged phospholipids, such as PS (Piccin et al., 2007). MV are purified from platelet-free plasma (PFP) samples, or can be detected in culture supernatant (Pankoui Mfonkeu et al., 2010). They are then incubated with annexin V, allowing flow cytometric analysis or enzyme-linked immunosorbent assays (ELISA) to detect the annexin V, and thus positive MV. However, this method has some shortcomings. Annexin V not only binds to MV, but also negatively charged phospholipids on cell fragments, which impacts on the accuracy of analysis of MV populations. In addition to this, it has recently been found that not all MV populations are annexin V-positive (Piccin et al., 2007).
Another method of MV detection involves labelling with antibodies. As MV express antigens on their surface when in an activated state, or when the cell is undergoing apoptosis, these can be targeted. Antibodies can specifically target surface markers indicative of particular cellular origins of MV, for example CD34, CD51, and E- and P-selectin for endothelial-derived MV (EMV), and CD31, CD41a, CD61, and CD63 for platelet-derived MV (PMV) (Piccin et al., 2007). Therefore, antibodies raised against these markers will detect EMV or PMV, respectively. Markers for other types of MV include CD11b for monocytes, CD3 for lymphocytes, and glycophorin A for RBC.
Due to inconsistent nomenclature use, microvesicles have been confused with other extracellular vesicles in the past, particularly exosomes and apoptotic bodies, however their size, contents, mechanism of formation, and membrane composition are extremely heterogeneous. While microvesicles, as previously described in section 3.1-4, are in the 200-1000 nm size range, exosomes are smaller, ranging from 40-200 nm, while apoptotic bodies are larger. In addition, these three types of vesicles are formed in very different ways, from different cellular compartments. Apoptotic bodies, like microvesicles, are plasma membrane-derived, however they form specifically in apoptosis-mediated conditions. Conversely, exosomes are formed within the cell, in endosomes or multi-vesicular bodies (MVB), and are released through a process called exocytosis, where the MVB fuses with the plasma membrane. Exosomes contain, as well as proteins and RNA from their cell of origin, a common set of evolutionarily-conserved protein molecules and double-stranded (ds) DNA (Thakur et al., 2014; Maguire, 2016), and are usually characterised by their surface markers. Microvesicles and exosomes are isolated via differing centrifugation protocols, but both play important immunomodulatory roles (Théry et al., 2002).
Low levels of MV can be detected in the circulation of healthy individuals, mostly of platelet origin, but also RBC, leucocyte and EC-derived (Piccin et al., 2007). It is known that MV are not simply inert elements, but instead are instead active vectors capable of stimulating a number of biological processes, including activation of coagulation, modulation of vascular function and induction of inflammatory processes (van der Heyde et al., 2006; Piccin et al., 2007; Wassmer et al., 2011a). The aminophopholipids present on the surface of MV provide binding sites for various clotting factors, including IXa, VIII, Va, prothrombinase and tenase. This allows them to play an active role in cell-cell interactions, propagation of signals, inflammation, coagulation, and vascular function (Combes et al., 2010).
In addition to the aminophospholipids, MV have other elements that play a role in their coagulating and physiological properties, thus contributing to haemostasis. EMV express ultra large vWF multimers, which stimulate platelet aggregation (Piccin et al., 2007), PMV express P-selectin (Freyssinet, 2003), and monocyte-derived MV (MMV) express TF and P-selectin glycoprotein ligand-1, which promotes intravascular thrombus formation (Celi et al., 2004). The expression of PS by MV is thought to contribute to the phagocytosis of apoptotic cells. The interaction of the PS receptor on macrophages and the PS expressed on MV has been examined, and thought to assist in the clearance of apoptotic cells (Piccin et al., 2007). In general, MV under normal physiological conditions play a very important role in homeostasis throughout the body.
Recently, the pathogenic role that MV play in several diseases has been highlighted. While MV play a role in maintaining homeostasis at normal physiological levels, as stated above, imbalances in numbers of circulating MV, have been shown to be associated with pathological conditions (Combes et al., 2010), playing both beneficial and detrimental roles. In individuals suffering from particular pathologies, Specifically, MV levels differ from the baseline concentration found in their healthy counterparts, either showing a significant increase or decrease. This change indicates their potential involvement in disease development, progression, or resolution in a wide range of infectious, autoimmune, cardiovascular, and inflammatory diseases, as well as a variety of cancers (Schindler et al., 2014).
Circulating MV numbers are decreased in several conditions, including systemic lupus erythematosus, Scott’s syndrome, and Burkitt’s leukaemia. Specifically, circulating MMV are decreased in active neuropsychiatric SLE (Crookston et al., 2013). Scott’s syndrome is a severe bleeding disorder in which microvesiculation is impaired (due to impaired phospholipid scramblase activity and thus, reduced PS exposure), leading to a reduction in the procoagulant effect of platelets and a decrease in the release of procoagulant vesicles (Toti et al., 1996).
Conversely, increased MV numbers have been detected in, and implicated in the development of many conditions, particularly those involving inflammation, coagulation, and altered vascular function (Coltel et al., 2006). Increased MV numbers have been demonstrated in a range of cancers, including in brain, breast, colorectal, lung and prostate cancers (Muralidharan-Chari et al., 2010; Thaler et al., 2012). Increased MV have also been implicated in several inflammatory pathologies, including atherosclerosis, diabetes mellitus, Crohn’s disease, and rheumatoid arthritis (Knijff-Dutmer et al., 2002; Leonetti et al., 2013; Alexandru et al., 2016). Indeed, PMV are known to be pro-thrombotic, and are increased in several cardiovascular pathologies, where they also function to activate EC (Zwaal and Schroit, 1997). EMV have been described as markers of endothelial dysfunction in multiple sclerosis, hypertension, and diabetes (Nomura et al., 2002; Horstman et al., 2004; Combes et al., 2005b; Muralidharan-Chari et al., 2010; Alexandru et al., 2016; Zinger et al., 2016). As well as multiple sclerosis, raised levels of MV have been identified in other neuropathologies, such as Alzheimer’s disease and, in the case of our study, cerebral malaria (Combes et al., 2005b; Xue et al., 2012).
Increased numbers of circulating MV are associated with several pathological conditions, including CM. MV are increased in both malaria patients and infected mice specifically with CM (Combes et al., 2004; Combes et al., 2005b; Pankoui Mfonkeu et al., 2010). TNF acts as an agonist in this process, circulating in high titres in severe malaria patients (Grau et al., 1989), and triggering a substantial increase in EMV displaying a pathogenic phenotype, by expressing coagulation and cell adhesion molecules (Combes et al., 1999; Combes et al., 2004; Wassmer et al., 2011b; Sahu et al., 2013). These EMV levels are increased in CM patients during the acute CM phase, compared to non-infected subjects and severe malarial anaemia patients (Combes et al., 2004). EMV levels were also shown to correlate with plasma TNF levels, suggesting that TNF could have a significant effect on vesiculation in vivo, similar to what was previously demonstrated in vitro (Combes et al., 1999).
In CM patients, PMV are the most abundant and their levels are significantly correlated with coma depth, thrombocytopenia, and disease severity (Piguet et al., 2002; van der Heyde et al., 2005; Pankoui Mfonkeu et al., 2010) (Weatherall et al., 2002; van der Heyde et al., 2006; Vlachou et al., 2006; Nantakomol et al., 2011). MV numbers were increased in CM patients, compared to trauma patients or those severely ill due to sepsis (Nantakomol et al., 2011). In vitro studies have also shown that PMV can transfer antigens to the pRBC membrane following adhesion, modifying their phenotype and dramatically increasing pRBC cytoadherence to EC, potentially leading to CM pathology (Faille et al., 2009b). These studies suggest a pathogenic role of MV in human CM.
Studies using the murine model of CM have provided more evidence for this pathogenic role. Plasma MV are increased in mice with CM, and specifically at the time of onset of neurological signs of the disease (El-Assaad et al., 2014). Further to this, adoptively transferring MV from mice with CM into healthy or infected recipient mice resulted in the arrest of MV in the vessels of infected mice only (El-Assaad et al., 2014). Conversely, mice with the ABCA-1 gene deficient do not up-regulate MV numbers and are 100% resistant to CM (Combes et al., 2005b). Pharmacologically inactivating the ABCA1 gene with pantethine had the same effect (Penet et al., 2008), and increased survival was observed in mice treated with anti-CD41 or 61 monoclonal antibodies (van der Heyde et al., 2005). MV from infected but not naïve mice induce potent activation of macrophages as measured by CD40 up-regulation and TNF production. However, similar levels of immunogenic MV were produced in WT and in TNF-/-, IFN-/-, IL-12-/-, and RAG-1-/- PbA-infected mice, but were not produced in mice injected with LPS (Couper et al., 2010). From this, they concluded that inflammation is not required for MV production during malaria infection and that, instead, pRBC-derived MV are the major inducer of systemic inflammation during malaria infection, raising questions about their role in severe disease and in the generation of adaptive immune responses.
Recent research has explored the possible presence of micro RNA (miRNA) in MV. miRNA are small, single-stranded, and highly conserved non-coding RNAs that regulate the translation of mRNA and protein. miRNA are found in plants, animals, and even some viruses, and control more than 30% of protein-coding genes, through post-transcriptional regulation of targeted gene expression and RNA silencing (Bartel, 2004). Particularly, in humans, ~2600 miRNA may be encoded, controlling around 60% of protein-coding genes (Griffiths-Jones et al., 2006; Friedman et al., 2009). This number is slightly lower in mice, with ~1900 mature miRNA identified. While the majority of miRNA are intracellularly located, some miRNA have been identified circulating in various biological fluids and cell culture media, termed extracellular miRNA (Turchinovich et al., 2013).
miRNA biogenesis has been reviewed a number of times (Bartel, 2004; Treiber et al., 2012). To summarise, miRNA genes are transcribed in the nucleus by RNA polymerase II (Pol II), to form primary miRNA transcripts (pri-miRNA), which contain 1-6 miRNA precursors, can be several thousand nucleotides in length, and are polyadenylated with multiple adenosines (a poly(A) tail). This transcript is spliced by the Drosha or Pasha enzyme to form precursor miRNA (pre-miRNA, ~70 nt). The pre-miRNA moves to the cytosol mediated by Exportin-5, and incorporated as single-stranded RNA sequences into the RNA-induced silencing complex (RISC), which contains the Dicer enzyme, and the Argonaute (Ago) protein family. pre-miRNA are then cleaved by Dicer to form mature effective miRNA (~22 nt), bound to Ago proteins within the RISC, and directed toward their target mRNA to be regulated. Mature miRNA are able to pair with mRNA perfectly or imperfectly, either inhibiting mRNA, or leading to their degradation, affecting downstream protein synthesis (Treiber et al., 2012). There is quite a lot of redundancy in miRNA targeting, whereby multiple miRNA can regulate the expression of a single mRNA or, conversely, one miRNA can target multiple mRNA, and subsequently affect several families of genes (Baltimore et al., 2008).
A number of methods are available to study miRNA interactions, biogenesis, expression, and function, including microarrays, quantitative real-time PCR, in situ hybridisation, and RNA sequencing (Thomson et al., 2007; Reid et al., 2011; Pritchard et al., 2012). These techniques can be applied to various sample types, as miRNA are found in blood, plasma, and serum, as well as other biological fluids such as urine, saliva, and other sample types such as tissues, model organisms, and host extracellular vesicles including exosomes and MV. miRNA profiling then forms the basis for further downstream target identification and analysis (Weber et al., 2010; Reid et al., 2011).
miRNA are known to play key regulatory roles in numerous biological processes, including cell proliferation, development, differentiation, and apoptosis (Bushati and Cohen, 2007; Liang et al., 2013), but have been shown to be dysregulated in a range of diseases caused by viruses, bacteria, and parasites (Ding and Voinnet, 2007; Hakimi and Cannella, 2011; Eulalio et al., 2012). miRNA have also been shown to play a critical role in regulating the cellular and molecular networks controlling the inflammatory process within a range acute and chronic inflammatory pathologies, including multiple sclerosis, rheumatoid arthritis, systemic lupus erythematosus, diabetes, inflammatory bowel disease, sepsis, and various types of cancer, during which their levels are altered (Fulci et al., 2010; O’Connell et al., 2012; Hessvik et al., 2013; Fourie et al., 2014; Tacke et al., 2014; Alexandru et al., 2016; Yang et al., 2017a). Indeed, there is a strong correlation between loss of enhanced expression of miRNA, and cancer (Hunter et al., 2008).
This range of disease areas into which miRNA research has expanded, has demonstrated the usefulness in profiling miRNA as biomarkers of severity; aiding diagnosis and prognosis (Reid et al., 2011). miRNA are generally well-suited as disease biomarkers, due to their stability in biofluids and widespread presence across biological sample types. Particularly, their presence in biological fluids such as blood, saliva, and urine allow them to be sampled relatively non-invasively (Pritchard et al., 2012; Xu et al., 2013). Therefore, miRNA have been targeted for therapeutic purposes, either using anti-miRNA to inhibit or reduce their expression, or miRNA mimics to verify and enhance their function, both for downstream effects in gene regulation (Hoss et al., 2015; Krützfeldt, 2016; Rupaimoole and Slack, 2017).
Few studies exist the examine the role of miRNA in malaria, however these have been reviewed in Chapter 2, and this has been explored further in Chapter 4. Fewer still explore miRNA within extracellular vesicles during malaria – to the best of our knowledge there are currently only three studies, one of which explores microvesicles specifically, instead of exosomes, and none examine cerebral malaria specifically, yielding the purpose of the study detailed in Chapter 3.
Previously, it was not known whether MV contained miRNA, but in the study by Hunter et al. in 2008, miRNA expression was identified and defined in circulating plasma MV. This was also found by Yuan et al. in 2009, where MV were engineered to express green fluorescent protein (GFP) and transfer of GFP and a subset of miRNA was observed in vitro (Yuan et al., 2009). Collino et al. found evidence of MV released from human bone marrow-derived mesenchymal stem cells and liver resident stem cells shuttling functional mRNA that were also found to contain selected miRNA. Further to this, those MV with highly expressed miRNA were transferred to target cells, allowing for the possibility that the biological effect of stem cells may, in part, depend on MV-shuttled miRNA (Collino et al., 2010). Analysing the miRNA and mRNA content of MV may hold clues to further understand their role in pathologies such as malaria.
The role of miRNA within MV and exosomes during malaria infection is just beginning to be explored. Particular miRNA, including miR-16, 17, 332, 451 and 497, are increased in exosomes from pRBC (Mantel et al., 2016; Yang et al., 2017b), and function to impact angiogenesis in mice, while also inhibiting tumour growth (Yang et al., 2017b). Furthermore, exosomes have been shown to contain host erythrocyte-derived miRNA-Argonaute 2 complexes, capable of modulating barrier function during malaria infection (Mantel et al., 2016). MV numbers are increased in pregnant women with placental malaria, in which total as well as specifically trophoblast-derived MV have been shown to play an immunopathogenic role (Moro et al., 2016).Current treatments and preventative measures of cerebral malaria.
Upon admission, most patients are given supportive therapy to improve their survival chances, and allow the therapeutic drugs time to work. Shock, hypoglycaemia, hypoxia, and severe metabolic acidosis are managed by saline, fluid and glucose resuscitation and by whole blood transfusions (Idro et al., 2005). Antibiotics can be administered prophylactically, to avoid possible secondary infections. Adults presenting with pulmonary oedema or renal failure can be treated with ventilation and renal dialysis, respectively (Greenwood et al., 2005).
Antimalarial drugs also exist to treat uncomplicated falciparum malaria. The cinchona alkaloid group, including chloroquine, takes effect during the later stages of parasite development (Idro et al., 2005). Cinchona alkaloids can be administered intravenously or intramuscularly, but are no longer useful in the treatment of most cases of malaria, due to the development of resistance (Conway, 2007). Sulfadoxine-pyrimethamine (SP) has been introduced to replace chloroquine, however resistance to this drug has also developed in most countries (WHO, 2016).
To avoid resistance developing and overcome the emergence of resistant parasites, combination drug therapies are now being utilised as first-line treatment for P. falciparum infections. This involves using multiple drugs with different modes of action and therapeutic targets but with similar half-lives (WHO, 2016). Artemisinin-based combination treatments (ACTs) have been effective in Africa and Asia. Artemisinins are active at early and late stages of parasite development (Idro et al., 2005) and ACTs represent the most effective treatment currently available, with recovery from malarial infection usually occurring within three days. ACTs are easier to administer than cinchona alkaloid derivatives and have fewer side effects (Idro et al., 2005). However, they cost up to ten times that of single drug therapies such as chloroquine (WHO, 2016). This increased cost, combined with the shortage of Artemisia annua (the plant source of artemisinin) has prompted production of synthetic versions of artemisinins (Greenwood et al., 2005). There is also increasing resistance to artemisinin in South-East Asia. Evidence of P. falciparum parasites with reduced in vivo susceptibility to artemisinin derivatives has emerged in western Cambodia (Noedl et al., 2008; Dondorp et al., 2009). However, prompt administration of ACTs for uncomplicated malaria patients resulted in a 99% reduced mortality in children ages 1-23 months, and by 97% in children aged 24-59 months (WHO, 2016). Parenteral artesunate or another artemisinin derivative is the recommended treatment for severe malaria, particularly CM, combined with early hospitalisation and critical care (Kyu and Fernández, 2009; Sinclair et al., 2012). Finally, a new class of antimalarial drugs – spiroindolones – has emerged, successfully inhibiting protein synthesis in the parasite (Rottmann et al., 2010).
Vector control has been one of the main efforts targeted at malaria eradication. Many approaches have been implemented, including the use of dichloro-diphenyl-trichloroethane (DDT) as an insecticide for indoor residual use, insecticide treated bed nets, draining of breeding sites and the use of lavivorous fish to control mosquitoes at the larval stages (Sachs and Malaney, 2002; WHO, 2011; WHO, 2016). Sterile insect technique is emerging as a potential mosquito control method. This involves transgenic, or genetically modified mosquitoes that have been made malaria-resistant. This involves introducing a gene into the mosquito that functions to impair the development of the parasite within its gut (Marrelli et al., 2007; Corby-Harris et al., 2010). In experimental malaria, promising fitness comparison results have been observed, providing potential implications for the control of mosquito vectors. Alternatively, new genetic techniques have been used to instead impair the ability of the mosquito to transmit the parasite, without affecting the mosquitoes otherwise – transgenic mosquitoes express antiparasitic genes introduced in their midgut epithelium, or bacterial species introduced into their flora that serve the same purpose (Ito et al., 2002; Riehle et al., 2007).
Currently, a completely effective vaccine does not exist, because of the complexity of parasite biology. Target antigens need to serve a function critical to the parasite, and be associated with naturally acquired immunity or be protective in animal models (Greenwood et al., 2005). So far, most efforts have been directed at the development of pre-erythrocytic stage vaccines designed to prevent the invasion of hepatocytes by sporozoites or to destroy infected hepatocytes. Currently, a single candidate vaccine has completed stage 3 clinical trials, and was approved for pilot trials in a few select countries by the European Medicines Authority (Committee for Medicinal Products for Human Use (CHMP), 2015). Another 22 projects are in phase 1 or phase 2 clinical trials, tackling the pre-erythrocytic, blood-stage and sexual-stage of the parasite, as well as testing the viability or irradiated sporozoites as an inactivated vaccine (WHO, 2016; WHO, 2017). RTSS/AS02A is the most advanced pre-erythrocytic vaccine, and is a hybrid consisting of a circumsporozoite protein of P. falciparum expressed with hepatitis-B surface antigen in yeast. This vaccine is given with the complex adjuvant, AS02, and has provided substantial, short-lived protection in volunteers exposed experimentally or naturally to bites from infected mosquitoes. This vaccine reduced clinical incidence of malaria by 39% overall, and specifically severe malaria by 31.5% (Greenwood et al., 2005).
Another type of vaccine that has the potential to help control the spread of malaria is a recombinant set of vaccines called transmission-blocking vaccines. These vaccines work by preventing the development of malarial parasites within the mosquito vector, and thus eliminating the cascade of secondary infections in humans. As a crucial stage of the Plasmodium life cycle is the evolution of the parasite in the midgut of the Anopheles mosquito, targeting midgut antigens that serve as ligands for the parasite offers a potential solution in the control of this disease (Mathias et al., 2012). For example, the highly conserved, midgut-specific, anopheline alanyl aminopeptidase N (AnAPN1) is a putative ligand for P. falciparum and vivax ookinetes invasion, and can be purified from an Escherichia coli vector. AnAPN1 is able to trigger the generation of functional malaria parasite transmission-blocking antibodies, and adsorbs completely to incomplete Freund’s adjuvant, a safe adjuvant designed to enhance the immune response of the recipient. The success thus far in the research suggests that this vaccine could be suitable to progress to phase 1 clinical trials (Mathias et al., 2012).