Typically illustrating the ‘manipulation hypothesis’, Toxoplasma gondii is widely known to trigger sustainable behavioural changes during chronic infection of intermediate hosts to enhance transmission to its feline definitive hosts, ensuring survival and dissemination. During the chronic stage of infection in rodents, a variety of neurological dysfunctions have been unravelled and correlated with the loss of cat fear, among other phenotypic impacts. However, the underlying neurological alteration(s) driving these behavioural modifications is only partially understood, which makes it difficult to draw more than a correlation between T. gondii infection and changes in brain homeostasis. Moreover, it is barely known which among the brain regions governing fear and stress responses are preferentially affected during T. gondii infection. Studies aiming at an in-depth dissection of underlying molecular mechanisms occurring at the host and parasite levels will be discussed in this review. Addressing this reminiscent topic in the light of recent technical progress and new discoveries regarding fear response, olfaction and neuromodulator mechanisms could contribute to a better understanding of this complex host–parasite interaction.
An equilibrated endoparasitic relationship results from a fine-tuned balance between host survival and parasite spreading. Toxoplasma gondii is an obligate intracellular protozoan parasite infecting a broad range of intermediate hosts that include virtually all warm-blooded animals. Prevalence is estimated at 30% in humans, but may surpass 60% in mice, rats and wild birds [1,2]. The parasite undergoes sexual reproduction exclusively in the intestine of its definitive hosts, members of the cat family (Felidae) leading to the shedding of highly infectious oocysts in the faeces. In the intermediate hosts, either oocysts (containing sporozoites) or tissue cysts (containing bradyzoites) coming from undercooked contaminated meat  differentiate into rapidly dividing tachyzoites that spread throughout the body. This acute phase of infection is most often asymptomatic and readily controlled by the immune system. Subsequently, the parasite differentiates into a slow-growing bradyzoite and forms cysts found predominantly in the brain, and also in skeletal and cardiac muscles, persisting potentially for the host's lifetime (Figure 1). It is important to consider that Toxoplasma epidemiology is in constant flux, sensitive to external factors, such as food habits of human consumers, management of livestock, hygienic standards of abattoirs, density of cats or wild felines in the environment and influence of urbanization on mammalian wildlife species .
The complex host interactions of Toxoplasma gondii.
Acquisition of genetic diversity being advantageous for every living organism, the goal of T. gondii is to reach its feline definitive host, achieve sexual replication and ensure successful transmission. Accordingly, it seems plausible that the parasite has established strategies to favour the meeting of the predator with its natural prey. In that respect, feline's preys can be considered as natural hosts and humans as incidental hosts for T. gondii. Consequently, parasites efforts to modulate the infected host behaviour might be partially unproductive apart from the natural host context.
The aim of this review is to recapitulate the behavioural dysfunction phenotypes associated with T. gondii infection in rodents and to relate them to the molecular mechanisms occurring at the level of host–parasite interplay during infection and brain colonization.
Infection routes, inflammatory response and clearance
The acute phase of T. gondii infection starts immediately after parasite ingestion, with tachyzoites spreading rapidly in epitheliums and reaching the blood stream. To mimic the natural infection mode in the mouse model, the most commonly used methods are either peroral or intraperitoneal administrations of a low number of cysts or tachyzoites, respectively. A key feature of T. gondii infection is the strong pro-inflammatory Th1 response induced and orchestrated by cells from the haematopoietic system such as granulocytes, macrophages, dendritic cells and lymphocytes. As part of the protective immune response, mouse cells rely on the activity of a subset of guanylate-binding proteins called immunity-related GTPases that are recruited at the PVM, resulting in the vesiculation and stripping of the parasitophorous vacuole membrane .
Although the mechanism explaining T. gondii access to the brain is still debated, two major scenarios emerge: the ‘Trojan horse’ hypothesis suggests that the parasite utilizes circulating infected cells to disseminate from sites of infection to the brain , and alternatively, free parasites in the blood compromise the blood–brain barrier (BBB) by invading, replicating in and lysing endothelial cells . Tachyzoites have been shown to exploit the migratory properties of dendritic cells (DCs) to spread throughout the host organism . Infected DCs exhibit a hypermigratory phenotype, due to the expression of the chemokine receptor CCR7, without DC activation . This phenotype is lost in the aspartyl protease 5 (ASP5) knockout parasites that are defective in the export of dense granule proteins, acting as effectors into host cells . T. gondii successfully crosses the BBB, ending up in cerebral microvessels, and subsequently establishing tissue cysts, predominantly in neurones. In healthy individuals, chronic infection has little to no clinical manifestations. However, immunocompromised infected hosts can experience reactivation of dormant bradyzoites into proliferating tachyzoites, leading to severe tissue damage. Therapies available to date, primarily pyrimethamine combined with sulfadiazine, eliminate circulating tachyzoites, but fail to eradicate the encysted bradyzoites. The persistence of chronic infection in the population also depicts a lack of sustained and effective antiparasitic immunity. According to the current state of knowledge, once a chronic infection is established, complete parasite clearance cannot be achieved.
Parasite-induced behaviour in infected hosts
Beyond the clinical symptoms: pathologies linked to toxoplasmosis
Although the chronic phase is largely considered asymptomatic, changes in behaviour have been associated with T. gondii infection in humans. Correlations have been shown between seroprevalence of T. gondii and schizophrenia, risk of traffic accidents, extraversion, less conscientiousness, personality disorders, bipolar disorder, obsessive compulsive disorders, Parkinson's disease, Tourette's syndrome, autism spectrum disorders, suicide, suicide in post-menopausal women and in patients with recurrent mood disorders . This extensive list illustrates the difficulty in defining a clear pattern of syndromes directly associated with toxoplasmosis.
A possible underlying mechanism explaining the effects on human behaviour was suggested to result from the stimulation of host dopamine pathways in the brain during T. gondii infection [11–14]. Dopamine functions as a neurotransmitter in the brain and plays a major role in several medical conditions. For instance, the imbalance of dopamine between the mesolimbic and mesocortical regions is suspected to participate in the development of schizophrenia . Interestingly, T. gondii contains two genes encoding aromatic amino acid hydroxylases (TgAAH1 and TgAAH2) that produce l-DOPA, the dopamine precursor. In consequence, the parasite metabolism could contribute to increased dopamine levels . However, parasites depleted in TgAAH1 and/or TgAAH2 did not have an impact on host dopamine levels, neither in vitro nor in vivo . Instead, the phenotypes were a severe defect in intestinal infection in the cats and a decreased yield of oocysts , distinctly unrelated to the manipulation of host behaviour. Furthermore, in this study, there was no measurement of change in global dopamine level in the brain , contrastive to a previous report . Although localized changes are not excluded, T. gondii infection did not result in a global change in dopamine level in the brain , suggesting that the behavioural anomalies are caused by alternative mechanisms.
Symptoms specific to rodents
The common symptoms associated with acute toxoplasmosis in rodents are lethargy, ruffled fur, hunched posture and transient reduction in body weight. They are typically associated with an inflammation phase that spans over 1–3 weeks following infection. Once clearance of circulating tachyzoites is achieved, the infection enters a chronic phase, with no symptoms display. Since the 1970s, numerous studies highlighted that T. gondii alters subtly the cognitive perception of chronically infected rodents in the face of predation risk, while the general health status of animals remains unaltered [2,19–21]. This hypothesis was conceptualized as ‘fatal feline attraction’  and served as foundation to future research. However, these findings were not always consistent across all studies and this appealing dogma was reappraised . Behaviours not obviously related to enhanced predation (e.g. social interactions and mate choice) also seem affected.
Impacts of Toxoplasma strains on the host during infection
The majority of T. gondii clinical isolates from North America and Europe belong to one of only three clonal lineages referred to as types I, II and III . Type I strains are relatively rare and characterized by a high level of acute virulence in the mouse model. Independent of the laboratory mouse strain, type I strains show a 100% lethal dose (LD100) for a single organism inoculation . Type II strains are most commonly associated with human infection, whereas type III strains are relatively common in domestic and wild animals and yet rarely found in human infection . Type II strains show intermediate virulence (LD50), whereas type III strains are almost avirulent . Importantly, studies on genetic diversity based on next-generation sequencing have revealed that the genomes of T. gondii type I, II and III strains are distinguished by the amplification of gene families that encode families of surface and secretory proteins vital for pathogenesis . Several of these secretory proteins play critical roles in virulence, while functions of the rest remain to be resolved.
Used in many behavioural studies, type II strains result in high tissue–cyst loads in the brain of mice and correspondingly cause high levels of immune-mediated brain inflammation [27,28]. Brain inflammation is measured by specific detection of known markers of inflammation, at the gene expression (qRT-PCR, microarray and RNAseq) or at the protein level (cytokine/chemokine array and imaging). Additionally, changes in the host transcription occur during T. gondii infection, such as alternation of microRNA expression [29–31]. Specifically, miR-146a and miR-155 are regulated by the transcription complex NF-κB and are modulated during T. gondii chronic infection, in a strain-specific manner . Variations in host gene expression in the frontal cortex have also been described during Toxoplasma infection . Markedly, the presence of the parasite alters the expression of the genes involved in the development of the forebrain, neurogenesis, sensory and motor coordination . Furthermore, infection leads to the down-regulation of olfactory receptors and dopamine receptor D4 . Notably, the effect on olfactory functions appeared to be more pronounced in male compared with female mice analyzed in the same study . Furthermore, epigenetic molecular mechanisms have previously been associated with long-term behavioural changes through active regulation of gene transcription in the central nervous system (CNS) .
Molecular determinants defining Toxoplasma-induced host manipulation
If manipulation of the host behaviour is a key determinant of the successful parasite transmission, this feature must be a general trait shared by many, if not all, T. gondii strains. A bulk of behavioural studies have preferentially used type II strains and showed effects during the early latent phase of T. gondii infection (e.g. 4–8 weeks post-infection) [22,34,35] as well as over longer duration, up to 7 months post-infection  (Figure 2). To evaluate the contribution of the early phase of acute infection, an attenuated type I strain and a low-virulent type III strain were used to infect and further test mouse aversion towards predator and non-predator urines . Interestingly, this experimental set-up revealed that mice infected with type I-attenuated parasites, presumably incapable of persisting as cysts in the brain, still exhibited attraction to cat odour months after complete clearance of infection . It is important to consider here that clearance of infection refers here to the absence of parasite DNA in the brain, as assessed by semi-quantitative PCR. Additionally, infection resulted in parasite and leucocyte infiltration of the brain region . These findings suggest that the three main lineages of T. gondii induce the loss of cat fear and point towards parasite replication in the peritoneal cavity and subsequent penetration in the mouse brain during acute infection being sufficient to trigger persistent loss of mouse innate aversion towards cat urine. Inflammation associated with brain colonization represents a decisive step for behaviour transformation, which certainly deserves further investigations.
Timing of behavioural tests with respect to the phase of infection.
Studies reporting behaviour alterations during T. gondii infection are typically performed more than 4–5 weeks post-infection, once a steady state is attained, to avoid interindividual differences during acute infection (Figure 2). The uneven nature of brain colonization during acute infection results in a non-polarized distribution, with tissue cysts found in all brain areas, as analyzed using standard histological techniques . However, if T. gondii appears to show no tropism for specific regions of the mouse brain, the degree of inflammation does not follow the same pattern. Specific regions, such as the prefrontal cortex, develop a more severe inflammatory response than other regions of the brain (Figure 1) . Although the level of IFN-γ is higher in the cortex than in the hippocampus or caudoputamen (assessed 52 days post-infection), there is no significant correlation between IFN-γ expression and the time spent freezing in the fear-conditioning test .
The immune-privileged nature of the CNS makes it vulnerable to any type of infection associated with the recruitment of immune cells. Representative of the transition from acute to chronic infection, complement C1q is an immune protein that bridges innate and adaptive immunity and clearing antigen–antibody immune complexes from systemic circulation . Compared with uninfected mice, cerebral C1q is up-regulated in response to latent T. gondii infection . Interestingly, previous reports showed that C1q is up-regulated in a range of brain disorders including Alzheimer's disease and schizophrenia [42,43]. In summary, T. gondii causes a moderate but sustained inflammation upon entry in the brain, which might result in local perturbation of brain homeostasis (for example, altered neurotransmitter levels) and have impact on the regulation of fear and anxiety.
Associated brain regions and neural circuitry
The random pattern of T. gondii distribution in the brain suggests an absence of selective tropism towards a specific cerebral region. Nevertheless, some studies reported higher cyst numbers in the olfactory bulb, the frontal cortex, and more importantly in the hippocampus, the amygdala and some areas of the midbrain involved in fear response, anxiety and olfaction . Of relevance, the amygdala, the hippocampus, the ventrolateral periaqueductal grey and the ventral anterior olfactory nucleus have been implicated in the expression of predator odour-induced unconditioned fear .
The amygdala is a critical region of the brain for the integration of cognitive functions, emotions and memory, the circuitry and role of which have been well conserved across evolution in higher eukaryotes. Anatomically, the amygdala is composed of multiple interconnected nuclei in the temporal lobe. The basolateral complex of the amygdala (BLA) is an important centre implicated in fear memory formation and anxiety response in the brain. The BLA processes and relays cognitive inputs to influence fear memory consolidation downstream in the hippocampus, caudoputamen and sensory areas . Amygdala circuitry and subsequent processes are regulated by stress hormones, such as glucocorticoids, that influence areas of the brain involved in emotional processing, learning and survival. Stress stimulations activate sympathetic and acute neuroendocrine responses that together induce a fight or flight type of response. The sympathetic activation results in epinephrine and norepinephrine secretion into the bloodstream, leading to an appropriate response to well-being or survival.
Once chronic infection with T. gondii is established in CNS tissues, parasite-induced mechanisms leading to neuronal dysfunction remain poorly understood. Whether T. gondii modulates directly or indirectly, memory association with fear conditioning remains an unresolved question. For instance, no significant deficit in fear memory of T. gondii-infected rats was observed in the fear-conditioning test . However, spatial memory is impaired in T. gondii-infected mice  as well as in rats . These observations were challenged by Ihara et al.  who showed that upon T. gondii infection, fear memory consolidation was impaired.
Among the regions controlling fear and stress responses, the periaqueductal gray (PAG) is known to be primary control centre for pain modulation and it is considered as an ‘analgesia centre’. The PAG is a grey matter area found in the midbrain, heavily interconnected with the hypothalamus and limbic forebrain structures, including the amygdala. The PAG is of considerable importance for organizing strategies to cope with intrinsic or extrinsic stress, and is also a site of action of analgesic agents including opioids or cannabinoids . Thus, the role of PAG is crucial from both a behavioural and therapeutic perspective . The PAG-mediated selective suppression of nociception (the sensory nervous system's response to harmful stimuli) causes the organism to respond in an appropriate manner to a life-threatening situation, without the counterproductive motor responses that might be provoked by a noxious input . Destabilization of the PAG during T. gondii brain colonization would be of considerable behavioural significance given the central role of the PAG in coordinating survival strategies.
Finally, the vomeronasal organ (VNO) or accessory olfactory is a sensory substructure of the olfactory system . VNO plays an essential role in the control of instinctive decisions determining animals’ social interactions, such as sexual attraction or on the opposite, avoiding aggressive type of behaviours . The crucial role of VNO in mouse is in both conspecific and interspecific recognition: it is initiating a robust response to predator cues and detect kairomones . Kairomones are mixtures of chemical substances emitted by an organism that mediate interspecific interactions. Specifically, the VNO is critical for predator odour-elicited defensive behaviours such as risk assessment or avoidance behaviour . Functionally, the VNO serves as an interface between the immune and nervous systems functions: it has been demonstrated that vomeronasal sensory neurone (VSN) detection of MHC class I peptides alters reproductive functions in female mice . This tends to indicate that VSNs contribute to the detection of information concerning immunological and genetic compatibility and possibly, infectious and genetic disease as well .
A persistent regenerative capability exists in the VNO epithelium and this capacity is preserved throughout life . In the context of a chronic infection, this feature is of importance and might participate in renewing cell population in the organ, while the parasite lyses host cells. T. gondii tissue cysts being dynamic but exhibiting a slow turnover , the VNO would represent a suitable environment to have an impact on predator-induced defensive behaviours in an epithelium with regenerative properties. To date, the presence of cysts in the VNO has not been investigated.
Are parasite effector molecules directly responsible for altering behaviour?
One implicit conclusion of the ‘manipulation hypothesis’ is that T. gondii actively modifies host behaviour to its advantage. Consequently, it should imply the secretion by the parasite of so-called manipulation factors or effector molecules that interfere with host brain functions. However, in the current state of knowledge, this claim cannot be formally established for two main reasons. Firstly, no tropism of T. gondii for brain region(s) associated with fear, emotion or stress response has been experimentally demonstrated. Considering that the parasite rather acts ‘on-site’, a certain degree of preferential localization in the brain would be expected. Secondly, given that the behavioural alterations persist over time, it would imply a constitutive secretion and sustained activity of the parasite ‘manipulation factors’. Even if tissue cysts are not completely quiescent, their metabolic activity is considerably reduced . Parasite proteins secreted by rhoptries or dense granules are plausible ‘manipulation factors’. However, while such secreted effector proteins are known to remodel the host transcriptome and signalling pathways during acute infection, little is known about what happens during the chronic stages.
Concluding remarks: Toxoplasma is not the unique manipulator among Apicomplexa
Ever since its postulation, the ‘fatal feline attraction’ theory has received considerable attention and further supported by several studies. However, studies on the manipulation of mouse behaviour during T. gondii infection are primarily descriptive with only a fractal understanding of what happens at the molecular level. In the absence of irrefutable proof linking the observed phenotypes with molecular mechanisms and parasite effectors, numerous questions remain unaddressed.
The modulation of host behaviour could be a fortuitous and indirect consequence of T. gondii infection. In this case, it is probable that other parasites from the same phylum that follow a similar modus operandi of infection would be responsible for comparable behavioural alterations. Hammondia hammondi is very closely related to T. gondii and represents an example of very restrictive, obligatory heteroxenous life cycle. Hammondia undergoes sexual development and oocyst formation in the cats, but is not transmissible between intermediate hosts . Another close relative of T. gondii, Neospora caninum, has diverged so that it performs its sexual part of the life cycle in Canidae. Neospora causes fatal loss in cattle and parallels that of T. gondii infection in sheep and goats. Both Hammondia and Neospora can be vertically transmitted , a trait shared by T. gondii in mice and humans . Examining the phenotypical impact of Neospora and Hammondia on infected mice in parallel to T. gondii would potentially help determine the nature of behavioural consequences towards certain predators.
Indeed, altering mice behaviour is not a specific feature of T. gondii among the apicomplexan parasites. Studies have revealed that mice infected with Eimeria vermiformis displayed a reduced avoidance to cat odour and a reduction in predator-induced fear when compared with uninfected controls [58,59]. Eimeria species are monoxenous, as they have a direct life cycle involving no intermediate hosts, and are capable of causing coccidiosis in animals such as cattle, poultry and smaller ruminants . Specifically, Kavaliers and colleagues showed that an acute subclinical infection with E. vermiformis attenuates the predator-induced analgesia . No symptoms at any stage of infection were observed in this study; this suggests that the altered responses of the infected mice to the cat were indirectly due to, or unassociated with, severe illness, and further supported a direct parasite-induced modification of host predator response. Intriguingly, male mice infected with E. vermiformis also displayed an increased preference for the odours of oestrous females, indicating that the parasitic infection does not selectively decrease olfactory sensitivity or locomotor activity . These observations raise the possibility that the reduction in predator avoidance may arise as a more general side effect of the host's responses to parasitic infection that facilitates interactions between infected and uninfected individuals and thus promotes parasite transmission.
These findings are surprising given that Eimeria is not a neurotropic parasite with the whole replication process taking place in epithelial cells of the gastrointestinal tract. Therefore, it is credible that the parasite-induced decreased analgesia is mediated through the enteric–CNS (ENS–CNS) cross-talk, also known as ‘gut–brain axis’ (GBA). The GBA is a complex communication system that not only ensures the maintenance of gastrointestinal homeostasis but probably also has an impact on higher cognitive functions and mechanisms such as immune activation.
The gastrointestinal location of the replicative cycle of Eimeria resembles the early phase of T. gondii acute infection following ingestion of sporulated oocysts and leading to rapid proliferation of tachyzoites in epithelial cells of the intestinal tract. Studies on the impact of T. gondii on the cellular composition of the colon in infected rats revealed that acute infection is associated with a decrease in the total number of neurones from the myenteric plexus . These findings indicate that T. gondii might cause quantitative and plastic alterations of fibres that connect either the CNS and ENS, or the CNS directly with the digestive tract. To date, little is known about the mechanisms by which T. gondii modulates the host ENS, the CNS–ENS cross-connections or the microbiome itself.
Reconsidering the ‘manipulation hypothesis’ taking advantage of recent progress in imaging and trying to correlate parasite replication in the digestive tract and brain colonization over time to behavioural tests could greatly help understand the mechanisms used by T. gondii to induce subtle and irreversible phenotypes in vivo. Moreover, the series of parasites mutants defective in exporting effector molecules will be of tremendous help towards the identification of the molecular mechanisms underlining subversion of host behaviour and regulator of host and parasite physiology.
Toxoplasma gondii infection induces behavioural alterations in mice, leading to the loss of cat fear.
Several studies have tackled experimentally this question, but the underlying neurological mechanism remains poorly understood.
In this context, the role of brain regions organizing fear, stress responses and olfaction certainly deserves further investigation.
In both acute and chronic stages of the infection, T. gondii actively remodels host transcriptome and triggers massive changes in the infected host cells.
Taken together, these observations lead to the speculative existence of a specific pathway involved in the alteration of host behaviour during T. gondii infection and possibly shared by other apicomplexan parasites.
basolateral complex of the amygdala
C-C chemokine receptor 7
central nervous system
enteric nervous system
parasitophorous vacuole membrane
vomeronasal sensory neurone
This work received funding from the Swiss National Science Foundation [310030B_166678] and the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme under Grant agreement no. 695596.
We thank Dr Glenn McConkey for critical reading of the manuscript.
The Authors declare that there are no competing interests associated with the manuscript.