African trypanosomes and brain infection – the unsolved question

I. INTRODUCTION

African trypanosomes comprise a group of flagellated protozoans of medical and agricultural importance as they infect humans or farm animals. In this review we concentrate on Trypanosoma brucei, a group of subspecies which infect humans causing sleeping sickness (T. b. gambiense and T. b. rhodesiense), or infect farm and game animals causing nagana (T. b. brucei). Today, more than 90% of human African trypanosomiasis (HAT) cases are induced by T. b. gambiense (the Western form), which causes a more chronic disease with a usual survival time of several years. T. b. rhodesiense, primarily located in Eastern and Southern Africa, is responsible for approximately 10% of cases and has a more acute progression, with death occurring within several months. Untreated, both forms of HAT are fatal. Following the bite of an infected tse-tse fly, the transmitted metacyclic insect form differentiates spontanously into a blood form of slender morphology that is completely covered by a densely packed glycoprotein coat. This protein belongs to a library of several hundred distinctly encoded genes that can be sequentially expressed, thus allowing antigenic variation (Marcello & Barry, 2007). The variant surface glycoprotein (VSG) coat protects the parasite against the host's cellular immune system and complement reactions. Antibodies are formed against the N-terminal part of each VSG type, allowing the infecting parasite carrying this specific VSG to be opsonized by antibodies and destroyed. However, since variants expressing different VSGs appear independently of the host's immune reactions, continuous fluctuation of parasite density in the blood results. In addition, high parasite density leads to a differentiation process and the formation of so-called ‘stumpy parasites’ which are unable to proliferate but are involved in tse-tse fly infection (Hamm et al., 1990).

Sleeping sickness was well known among African populations since ancient times as a cause of periodic devastating epidemics (Slane, 1968; Atkins, 1978). Trypanosomes were recognized as causative agents of the disease in 1903 (Davies, 1962), after the parasite was first detected in the blood of infected humans (Dutton, 1902). The observable symptoms, notably the sleep disorder, led to the assumption that trypanosomes invade the brain. This was experimentally confirmed in 1903, when T. b. rhodesiense was for the first time observed within the cerebrospinal fluid (CSF) (Castellani, 1903; Calwell, 1937).

Calwell (1937) first discussed involvement of the choroid plexus (CP), which is responsible for the production of CSF, in trypanomosiasis. Thus, this parasite was thought to be able to cross the blood–CSF barrier (BCB) (Fig. 1). By contrast, imaging of cerebral trypanosomiasis revealed perivascular cuffing within the white matter (substantia alba) of the brain (Enanga et al., 2002), gliosis (Poltera, Owor & Cox, 1977), haemorrhages (Kennedy, 2004), and subcortical perivascular demyelination (Enanga et al., 2002). In some cases, hyperintense signals of the white matter were found in T2-weighted magnetic resonance images, which were interpreted as oedema or bleeding (Spinazzola et al., 1989; Braakman et al., 2006). Furthermore, lesions within the grey matter, the cerebellum and the brain stem were found (Kager et al., 2009). Together, these findings suggest that trypanosomes can cross the blood–brain barrier (BBB), although infiltration via the Virchow–Robin space (see Section IV.1) cannot be excluded (Fig. 2). All these studies were limited by the fact that human brain samples can only be analysed post-mortem. An autopsy would have to take place immediately after death in order to detect trypanosomes in situ. This is especially difficult in endemic regions. Therefore, the route of infection has never been systematically followed using patient material, and publications showing histological sections with trypanosomes in the human brain are extremely rare (Poltera et al., 1977). Much effort has been made to investigate trypanosomiasis in animal models, especially rodents and monkeys.

Most animal studies using T. b. brucei (Masocha et al., 2004), T. b. rhodesiense (Bafort & Schmidt, 1983) or T. evansi (Tuntasuvan et al., 2000; Rodrigues et al., 2009) showed trypanosomes 21–63 days after infection within the brain of rats (Poltera et al., 1980; Schmidt & Bafort, 1985). To assist in the development of medications with minimal side effects, a crucial question is whether (and when) the parasite crosses the BCB or the BBB.

The focus of this review is on aspects essential for understanding the interaction of trypanosomes with the BBB and the BCB. We consider literature from a wide range of disciplines: areas from entomology to inflammation research, from developmental biology to public health, from pharmacology to protozoology, and from clinical neurology to sleep research.

II. SYMPTOMS OBSERVED IN SLEEPING SICKNESS

Sleeping sickness is transmitted by tse-tse flies, which belong to the genus Glossinae. Thus the endemic area is limited to the so-called tse-tse belt, i.e. sub-Saharan Africa. During the feeding process, blood-form trypanosomes are taken up by the fly and develop into a midgut form that enters the salivary gland, where it differentiates firstly into a cell-attached epimastigote form and finally to a free-swimming metacyclic form. During a subsequent blood meal, these trypanosomes enter the skin lesion with the release of anti-coagulatory saliva. After a brief period in the skin, the parasite may travel to blood and lymph vessels to begin the infection process (Caljon et al., 2016). A localized immune reaction at the site of the bite often leads to the formation of a chancre of about one inch in diameter that is associated with pruritus. Shortly after the bite, a typical first symptom of Western HAT is the Winterbottom's sign, i.e. a lymphadenopathy in the lateral cervical region of the neck.

Winterbottom's sign is usually absent in the haemolymphatic stage of Eastern HAT, which progresses more rapidly and often leads to pericardial effusion, pulmonary oedema and acute myocarditis (Fig. 3).

The second stage of infection, the so-called meningo-encephalitic stage, commences after a delay of weeks (Eastern HAT) or months (Western HAT). Since treatment and prognosis depend on the infection stage, the World Health Organization (WHO) recommends a lumbar puncture for diagnosis. Interestingly, trypanosomes do not survive long in CSF as it contains trypanocidal compounds (Pentreath, Owolabi & Doua, 1992; Wolburg et al., 2012). Thus, if parasites are found in the blood but not in the CSF, a cell count of more than five lymphocytes per microliter is considered positive for CSF infection. The appearance of trypanosomes in CSF is only indicative of parasites in the meningeal area and does not necessarily mean there is an infection of brain parenchyma (neuropil). The latency with which trypanosomes are detected in the central nervous system (CNS) varies considerably. A chronic stage with eventual exacerbation is now thought to exist after several case reports emerged describing patients who showed progression after more than a decade (reviewed in Sudarshi et al., 2014). From the CSF trypanosomes spread to susceptible tissues close to the ventricular system that do not have the BBB. This includes the circumventricular organs (CVOs), which are organized around the ventricles and possess fenestrated capillaries. Impairment of their physiological functions corresponds to commonly observed HAT symptoms. For example, the area postrema controls the emetic reflex and may become activated to cause nausea. Likewise, progressing somnolence and fragmented sleep patterns may be caused by deregulated melatonin secretion from the pineal gland. In addition, trypanosomes are able to produce and secrete prostaglandin D2 (PGD₂) (Figarella et al., 2005, 2006; Duszenko et al., 2006), which may be at least partially responsible for elevated PGD₂ levels recorded in late-stage HAT patients (Pentreath et al., 1990). PGD₂ induces sleep and may also be involved in the deregulation of sleep–wake cycles (Urade & Hayaishi, 2011). Infection of the neurohypophysis, the vascular organ of the lamina terminalis and the subfornical organ affects regulation of the secretion of vasopressin and luteinizing hormone releasing hormone (LHRH). Eventually, HAT patients show oedema as well as impotence and infertility (Kennedy, 2012). However, other observed symptoms such as gait ataxia and facial nerve paralysis await explanation.

Neurological and psychological symptoms are common in second-stage HAT but can be misdiagnosed as symptoms of depressive disorders. Personality changes, hallucinations and paranoid delusions coincide with encephalitis. Soon after observation of these symptoms, patients become less responsive and fall below a score of 12 on the Glasgow Coma Scale (GCS) (MacLean et al., 2012). Death usually occurs by heart failure following pancarditis, opportunistic infection, malnourishment or encephalitis. Interestingly, in experimental monkey infections, cases of early deaths and late deaths could be differentiated. Early deaths occurred primarily by heart failure, meningitis or other organ damage, while the latter was exclusively induced by encephalitis (Schmidt, 1983).

There is still no complete biological explanation of HAT symptoms (Bruzzone et al., 2009). Neurological disorders might be due to factors secreted by the parasite, or by the host's immune system or by mutual modulation (Kristensson et al., 2010). Sleep–wake deregulation has also been observed in T. b. brucei-infected rats (Grassi-Zucconi et al., 1995; Darsaud et al., 2004; Seke Etet et al., 2012). This may be caused by the presence of trypanosomes within the pineal gland and the secretion of sleep inducers like PGD₂ (Pentreath et al., 1990; Lundkvist, Kristensson & Bentivoglio, 2004; Figarella et al., 2006; Urade & Hayaishi, 2011) or tryptophol (Richards, 1965; Seed, Seed & Sechelski, 1978; Cornford et al., 1981). However, secondary effects on cytokine production also might lead to disruption of the circadian rhythm (Kristensson et al., 2010). African sleeping sickness causes changes to a number of hormones, including cortisol, prolactin and melatonin (Kristensson et al., 1998; Lundkvist et al., 2004). The photo-optic calibration of the sleep–wake rhythm takes place inside the suprachiasmatic nucleus. Neurotransmitters (especially glutamate) induce expression of the phosphoprotein c-fos. T. b. brucei-infected rats down-regulate glutamate receptors and thus have decreased levels of c-fos, maybe in response to an increase in levels of interferon gamma (IFNγ) (Lundkvist et al., 2004). In addition, changes to levels of tumor necrosis factor alpha (TNFα) also have an impact on the sleep–wake rhythm (Rodgers, 2010). Last but not least, cytokine-driven production of nitric oxide led not only to increased permeability of the BBB, but also to increased synthesis of prostaglandins (Enanga et al., 2002).

Other symptoms such as confusion, disturbed consciousness, coordination and sensory disorders could be the consequence of an increased autoimmune response (Galvao-Castro, Hochmann & Lambert, 1978). Antibodies against galactocerebrosides and neurofilaments circulate in CSF of HAT patients that may cause demyelination of nervous tissue (Enanga et al., 2002). However, all symptoms are reversible after eradication of the parasite by successful medical treatment. Thus it seems unlikely that trypanosomes have a direct cytotoxic effect on brain cells, as this would prevent full recovery. It has also been shown that trypanosomes do not have any impact on CNS cells during in vitro cultivation (Stoppini et al., 2000).

Attempts to link the appearance of different symptoms with progression of the disease have failed as published case reports are inconsistent (MacLean et al., 2010). Therefore, there is currently no clear symptom-based indication of the stage at which trypanosomes appear in CSF, when they move to the meninges and the CVOs and when they cross the BBB to invade the brain. We have attempted in Fig. 3 to link some prominent symptoms to different stages of the disease. It should be noted, that most symptoms are related to the function of the CVOs and meninges, while encephalopathy is clearly related to infiltration into the neuropil.

III. THE BRAIN STAGE IN SLEEPING SICKNESS

IV. THE MENINGEAL STAGE OF SLEEPING SICKNESS

(1) Structure and function of the BCB and the circumventricular organs

The mammalian brain contains four ventricles, i.e. spaces filled with CSF and separated from the brain parenchyma by a layer of ependymal cells. Each ventricle contains a choroid plexus (CP) that invades the ventricle and is responsible for CSF formation. For this purpose, the blood vessels of the CP have fenestrated endothelial cells which allow diffusion of serum from the blood vessel into the plexus stroma (Fig. 1). The outer layer of the CP is formed by epithelial cells, which are interconnected by TJs. On its way from stroma to ventricle space, CSF is formed from serum and is rather similar in terms of its osmolarity, water content and pH, while levels of glucose, amino acids, cations and anions are reduced by a factor of 1.5–2. Most strikingly, the protein concentration drops by a factor of about 200 (Lentner, 1981; Ernst, Palacios & Siegel, 1986; Lindsey et al., 1990). Thus, CSF is not a simple ultra-filtrate but is specifically modified, including the addition of neuropeptides.

The CP epithelial cells are sometimes referred to as ependymal cells since they are in continuity with the ependyma of the periventricular lining, but they are not identical. The main differences between ependymal and epithelial cells are: (i) the presence of a basal lamina in the CP, but not in the ependyma, and (ii) that the CP epithelium has barrier properties (continuous TJs), while the ependyma contains only adherens junctions and fragmented discontinuous TJs (Fig. 1).

The ependymal and epithelial cell layers meet at the neck of the plexus. Due to their special topology controlled fluid release is possible across the ependymal border, while cells cannot pass (Linß & Fanghänel, 1998). Thus, the plexus epithelial cells with its TJs form the BCB. In contrast to the endothelial BBB, the epithelial CP is characterized by high expression of at least claudin-1, -2 and -11, resulting in TJs formed by parallel arrangements of strands (Wolburg et al., 2001).

Transmigration of pathogenic bacteria across the plexus epithelium of the BCB has been found to be a para- or transcellular process (Wewer et al., 2011; Schwerk et al., 2012). Based on a daily CSF production of 500–700 ml in humans, the total volume of 120–200 ml of CSF is renewed 3–4 times a day. CSF formed in both lateral ventricles combines and flows via the intraventricular foramen of Monro to the third ventricle to join with CSF formed here. It then reaches the fourth ventricle via the aqueduct of Sylvius. The fourth ventricle in turn is connected via the foramen of Magendie with the cisterna magna (Fig. 4). From here CSF flushes into the subarachnoid space, i.e. the open space between the arachnoid mater and pia mater (Fig. 2B), which also extends into the vertebral canal. Below the skull, CSF is reabsorbed by arachnoid granulations. These are vessel-free protrusions of the arachnoid that reach superficial brain veins through gaps in the dura mater or the dura-encapsulated cranial sinus. Working like a valve, arachnoid granulations maintain a unidirectional flow whenever the intracranial pressure exceeds a defined threshold (Siegel, Agranoff & Albers, 1999). In addition, a direct connection between CSF and lymphatic vessels was recently discovered and is named the glymphatic system (Jessen et al., 2015).

Keeping the brain floating in CSF is necessary for several reasons: it serves to protect the tissues from suddenly arising acceleration or deceleration forces which could lead to severe trauma. Minimal volume changes (e.g. inflammatory swellings) of the brain can create life-threatening intracranial pressure increases. This is partially mitigated by controlled CSF reabsorption. In addition, CSF contributes to nutrient supply and clearance of metabolites by crossing the ependymal cell border. In this way, secreted substances can be easily transported. A role in signal transduction has also been discussed (Siegel et al., 1999; Jessen et al., 2015).

The CVOs are unpaired and located medial of the ventricle walls (Fig. 4). Except for the subcommissural organ, they do not possess a BBB, but are organized similarly to the BCB (Duvernoy & Risold, 2007). This is essential, as they secrete hormones or fulfil sensory tasks. The pineal gland, for example, releases the melatonin involved in defining the circadian rhythm. The eminentia mediana secretes neuropeptides, and the posterior pituitary produces antidiuretic hormone and oxytocin. Other CVOs have to react to substances circulating in the blood. The organum vasculosum of the lamina terminalis is involved in the regulation of thirst, hunger and fever. The sub-fornical organ reacts to angiotensin II in order to regulate electrolyte and water balance, and the vomiting centre (area postrema) needs to detect potential toxins. To meet these requirements, the CVOs have fenestrated capillaries, like those within the CP. The barrier to the adjacent brain parenchyma is formed by specialized ependymal cells, so-called tanycytes, which are interconnected by TJs that block cellular crossings.

V. IS THERE A CHRONIC MENINGEAL STAGE IN SLEEPING SICKNESS?

In HAT, detection of trypanosomes in CSF is a diagnostic tool for second-stage infection. Usually, finding trypanosomes in CSF is considered equivalent to the appearance of the parasite within the brain, hence the term ‘meningo-encephalitic’ stage. In principle, however, parasites crossing the BCB have no access to brain parenchyma, but spread throughout the subarachnoid space. Likewise, trypanosomes passing the equivalent of the BCB, i.e. the fenestrated blood vessels of CVOs, stay within these CVOs but cannot proceed to the brain parenchyma because tanycytes act as a physical barrier. As in CSF, trypanosomes within the CP seem unable to seed permanently within this organ as they lose their VSG coat and degrade (Mogk et al., 2014b). Instead, with each newly appearing population peak in the blood, trypanosomes recross the BCB and a new infection episode occurs, detectable by cyclic infection of CSF (Mogk et al., 2014b). In contrast to the fluctuating population in the plexus, the ventricles and the subarachnoid space, parasites are found permanently within the pia mater (Wolburg et al., 2012). Here they must be safe from any trypanocidal activity, most likely from neuropeptides (Gonzalez-Rey, Chorny & Delgado, 2006; Delgado et al., 2009), that is present in CSF. It is also likely that this acts as a source for trypanosomes reappearing in the blood. So-called ‘relapse’ is observed following trypanosome eradication from the blood by drugs such as suramin, which do not cross the BBB. By contrast, it is difficult to imagine how trypanosomes could reappear in blood starting from the neuropil.

There is no doubt that African trypanosomes are able both to cross the BCB and to appear within the neuropil; both facts have been well documented (see Sections III and IV). There is controversy, however, regarding when during an ongoing infection the parasite escapes from the haemolymphatic system. Generalized statements claim that it takes several months for T. b. gambiense and several weeks for T. b. rhodesiense to appear in the brain. This proposed delayed crossing of the BBB or BCB, however, has not been systematically established in natural human infections. If true, latency of CNS infection would certainly require a suitable explanation, but so far this is purely speculative. A plausible argument would be that antigen–antibody complexes are needed (Poltera et al., 1980; Lambert, Berney & Kazyumba, 1981), but an experimental approach in a rodent model could not confirm this (Wolburg et al., 2012). Instead, in rat infections periodic appearance of parasites in CSF, 1 day following a parasitic population peak in the blood, was found (Fig. 3B). In this case an additional mechanism to open the BCB is not necessary, as opening may depend on the appearance of short stumpy parasites, which express a surface-bound metalloprotease, major metalloprotease B (MSP-B; Grandgenett et al., 2007; de Sousa, Atouguia & Silva, 2010) to open the basal lamina surrounding the plexus blood vessels (Kleiner & Stetler-Stevenson, 1999). Artificial placement of bloodform trypanosomes into brain parenchyma never induced a brain infection; these parasites did not grow and died within a few days (Mogk et al., 2014b).

As mentioned above, several experimental approaches led to the observation that trypanosomes can cross the BBB, appearing in perivascular locations within the brain parenchyma (Masocha et al., 2004; Frevert et al., 2012). Interestingly, using a rather high parasite number injected intravenously into mice, trypanosomes appeared in perivascular locations just hours following injection (Frevert et al., 2012). Unfortunately, the fate of these parasites was not analysed over subsequent days and their ability to undergo cell division was not assessed. The question thus remains whether trypanosomes are able to survive and grow within the brain parenchyma for the several years of a natural T. b. gambiense infection. This is especially so for the ‘healthy carrier of trypanosomiasis’, i.e. asymptomatic humans infected with HAT that have been followed up for up to 15 years (Wurapa et al., 1984; Franco et al., 2014a), or in the case of an African who developed sleeping sickness symptoms almost 30 years after leaving Africa (Sudarshi et al., 2014). It is hard to believe that a symptomless brain infection was present for 30 years without causing neurological disabilities. Another argument against brain infection is that virtually no neurological symptoms remain after successful treatment. In this context, experiments performed on vervet monkeys are interesting. Following infection, two populations could be clearly distinguished: one group died rather early, usually because of heart failure, while the other group died later from encephalitis (Schmidt, 1983). This result suggests that trypanosomes may enter the brain only in a late phase of infection and that this encephalitic stage is inevitably fatal.

During the meningeal stage, a cell density-control mechanism must be in place to keep the parasite population within the available space, i.e. up to six pial cell layers. Since CSF has trypanocidal activity, parasites pushed out of the pia mater will be killed unless they can reach a blood vessel to induce a relapse. It is thus interesting to note that evidence exists suggesting a direct connection of CSF with the lymphatic system in the nasal pharynx (Zakharov, Papaiconomou & Johnston, 2004; Pollay, 2012). It is tempting to speculate that this glymphatic pathway is the route that trypanosomes use from the meninges to the haemolymphatic system during a relapse. This pathway seems also be important for other pathogens that invade the meninges and the brain, such as protozoan (e.g. Toxoplasma gondii causing toxoplasmosis), bacterial (e.g. Treponema pallidum causing syphilis) or viral (e.g. tick-borne encephalitis virus) infections. An increasing population density of trypanosomes induces differentiation from the slender to the stumpy morphology (Reuner et al., 1997; Seed & Black, 1997). Stumpy parasites produce PGD₂ (Figarella et al., 2005), which by a paracrine or autocrine mechanism leads to caspase-free apoptosis of stumpy cells (Duszenko et al., 2006). Since slender trypanosomes do not undergo apoptosis, persisting infection is ensured (Duszenko et al., 2006). This view is also supported by the observation that PGD₂ is markedly elevated in CSF of late-stage HAT patients (Pentreath et al., 1990). PGD₂ is also well known as a sleep inducer (Urade & Hayaishi, 2011) and this prostanoid of parasitic origin may thus contribute to the name-giving symptom of sleeping sickness.

VI. SYNOPSIS AND FUTURE PERSPECTIVES

It is surprising that after a period of more than 100 years of intense research on sleeping sickness and African trypanosomes, the exact details of the meningo-encephalitic stage are still elusive. Herein we discuss representative studies in an attempt to present a clear and comprehensive picture about what is known and what is still unknown regarding this stage.

There is consensus that trypanosomes are present within the meninges and within the brain parenchyma at a certain phases of infection. However, there is no conclusive evidence as to exactly when trypanosomes cross the BBB or the BCB. Since a defined mechanism explaining how trypanosomes cross either of these two barriers has not been provided, initiation of the second stage remains speculative. Statements that it takes weeks to months in rhodesiense infections and months to years in gambiense infections obviously derive either from speculations about the infection period based on when an individual had been in an endemic area, or from observations correlating appearance of a chancre with the first detection of typical HAT symptoms. However, this is not necessarily identical with the first appearance of parasites outside the haemolymphatic system. In fact, in a rodent model, it takes some 20 days for T. b. brucei to be detectable in brain after intraperitoneal infection (Poltera et al., 1980; Schmidt & Bafort, 1985); CSF was not examined in these experiments. In another study, trypanosomes in a rat infection appeared in CSF just 1 day after the first blood population peak and then periodically, linked to waves in blood infection (Mogk et al., 2014b). This result suggests that trypanosomes may appear in CSF at a very early point of infection to seed within the pia mater and also that crossing of the BCB depends on population density and thus, as already discussed, possibly on the formation of stumpy parasites.

Taking all the experimental evidence and medical observations into consideration, we propose that infection in sleeping sickness may include three phases. The first haemolymphatic stage starts shortly after the tsetse fly's bite, as the parasite has access to the blood and lymphatic systems. The second meningeal stage starts after the first sufficiently high blood population density peak and may depend on the appearance of short stumpy parasites to achieve entry across the BCB. The third encephalitic stage can be characterized by a breakdown of the brain's protective borders thus allowing mass invasion of trypanosomes to induce a fatal encephalopathy (Fig. 3B). This view is consistent with observations by Schmidt (1983) on artificially infected vervet monkeys. This model could explain several observations in natural infections: (i) trypanosomes cross the fenestrated endothelium as no additional mechanism is necessary besides high cell density allowing formation of stumpy parasites. (ii) Meningeal infection includes infection of CVOs and thus explains the onset of many symptoms observed during sleeping sickness. (iii) Because of controlled cell-density regulation of parasites within the pia mater by PGD₂-induced apoptosis, inflammation processes are prevented and a chronic phase may last for years or even decades. (iv) Relapses, regularly observed after clearance of parasites from blood, can be explained by reabsorption of CSF into the venous system by CSF–lymphatic bridges, while retrograde transfer across the BBB is difficult to imagine. (v) Full recovery after successful treatment of HAT is possible if parasites reside in the meninges only, since no brain structures would have been damaged. (vi) A non-life-threatening meningeal stage would suit transmission of the parasite, as it allows permanent blood infection even if parasitaemia is controlled by a successful immune response.

The model does not explain infection of the brain parenchyma, as observed in experimental and natural infections. Consistent with the above-described model, a meningeal infection could overwhelm control mechanisms and parasites infiltrate via the Virchow–Robin space to induce lethal encephalopathy (Fig. 3). This could happen during an acute attack, perhaps due to a weakened constitution of a receptive individual. Another scenario would include permanent crossing of the BBB by some trypanosomes which are then removed by e.g. microglia (H. Wolburg & M. Duszenko, unpublished observations), as observed in experimental intracerebral infection studies (Mogk et al., 2014b). In this case, lethal encephalopathy would occur after successful brain defences collapse either due to ongoing permanent infiltration of parasites or a generally weakened constitution. In contrast to these models, we consider a permanent and long-lasting infection of the brain parenchyma as rather unlikely, although we cannot exclude this possibility.

Despite common assumptions that the brain stage of HAT is well understood, questions still remain. The most important question is what molecular mechanism allows the trypanosome to move beyond the haemolymphatic system to infect the meninges and the CNS. Solving this will provide information on the length of time after an infected tse-tse fly bite that the parasite resides exclusively within the haemolymphytic system, and whether or not an explicit meningeal stage exists. Currently available drugs for second-stage HAT all have severe side effects, because they have been selected for their ability to cross the BBB. By flooding the brain, these drugs also affect delicate neural structures. If, however, a prolonged meningeal stage does exist, drugs should be selected for their ability to distribute within the meninges instead of the brain parenchyma. Such drugs would be in direct contact with the parasite and side effects thus may be minimized.

It is hoped that sleeping sickness will be eradicated or become manageable within the next decade, as suggested by the WHO (Franco et al., 2014b). However, with central questions about the infection mechanism unsolved, the disease may reappear at cyclic intervals as chronically infected individuals or animal reservoirs become infectious again.

VII. CONCLUSIONS