Cognitive Dysfunction in Mice Infected with Plasmodium berghei Strain ANKA

Abstract

Cerebral malaria complicated by cognitive sequelae is a major cause of morbidity in humans infected with Plasmodium falciparum. To model cognitive function after malaria, we created a rodent model of cerebral malaria by infecting C57BL/6 mice with Plasmodium berghei strain ANKA. After 7 days, an object-recognition test of working memory revealed a significant impairment in the visual memory of infected mice. This impairment was observed in the absence of confounding effects of infection. The cognitive dysfunction correlated with hemorrhage and inflammation. Furthermore, microglial activity and morphological changes detected throughout the brains of infected mice were absent from the brains of control mice, and this correlated with the measured cognitive defects. Similar testing methods in human studies could help identify subjects at risk for an adverse cognitive outcome. This murine model should facilitate the study of adjunctive methods to ameliorate adverse neurological outcomes in cerebral malaria.

Cerebral malaria due to Plasmodium falciparum infection is a major cause of morbidity and mortality in the developing world. It is estimated that 1 million children die each year of this disease in sub-Saharan Africa [1, 2]. Approximately 10%–28% of children who survive an episode of cerebral malaria develop neurological sequelae [1, 3, 4]. It has become increasingly recognized that cerebral malaria may cause persistent neurological and cognitive deficits that span a wide range of cognitive functions long after the infection has been successfully treated with antimalarial drugs, even in asymptomatic individuals [1–8]. Persistent deficiencies include attentional memory, learning and language impairments, visuospatial and motor deficits, and psychiatric disorders [3, 5, 6, 8–15].

The animal models of cerebral malaria faithfully recapitulate many of the characteristics of the human syndrome and these have been important in investigating pathogenesis of the disease [16]. For example, C57BL/6 mice infected with Plasmodium berghei strain ANKA exhibit neurological and behavioral impairments that are similar to those in humans infected with P. falciparum, including ataxia, hemiparaplegias, and seizures [16–18]. These neurological sequelae correlate with pathological findings in the brains of humans with cerebral malaria and of experimental animals [2, 5, 16, 17]. There have been few systematic studies of cognitive outcomes of cerebral malaria in animal models. However, damage to the brain in cerebral malaria occurs in regions known to be important in memory, such as the fornix, cortex, and hippocampus [19]. Astrogliosis and microglial activation in particular are found in these regions [20, 21] and precede neurological alterations [20]. In addition, several reports have documented increased levels of inflammatory mediators, such as cytokines and chemokines, in the brains of P. berghei ANKA–infected mice [21, 22]. In the present study, we demonstrate that mice infected with P. berghei ANKA display demonstrable cognitive impairment that is not observed in uninfected mice and that this impairment is associated with pathological findings in the brain.

MATERIALS AND METHODS

Mouse infection

Six-week-old C57BL/6 female mice (Jackson Laboratories) were infected with 5 × 10⁵ red blood cells parasitized with P. berghei ANKA diluted in 500 μL of PBS or were injected with uninfected blood diluted in PBS. The mice were then randomly separated into 2 groups (N=8). Parasitemia was monitored by examining Giemsa-stained blood smears on days 6 and 8 after infection. Experiments were performed in accordance with the Institutional Animal Care and Use Committee of the Albert Einstein College of Medicine.

Behavioral assays

The object-recognition test of working memory [23], which is based on the natural and robust tendency of rodents to preferentially explore novel objects, was performed as described elsewhere [23]. Briefly, in trial 1, mice were placed in an opaque acrylic arena (area, 97 cm²) and allowed to freely explore 2 identical, nontoxic objects (e.g., plastic, glass, or ceramic items) for 3 min. The time spent exploring the objects was recorded manually with timers, and locomotor activity was assessed simultaneously with Viewer tracking software (Biobserve), after which the mouse was returned to the home cage. Following a retention interval of 60 min, the mouse was returned to the arena, where one of the familiar objects had been replaced with a novel object (trial 2). The mouse was again allowed to explore for 3 min. Exploration of both the novel and familiar objects was given a preference score, defined as the time spent exploring the novel object divided by the total time spent exploring both objects, expressed as a percentage. A preference score of 50% indicated chance exploration (i.e., both novel and familiar objects are explored equally). Exploration of the objects was defined as any physical exploration (whisking, sniffing, rearing on, or touching the object) and approach and obvious orientation within 3 cm of the object.

Novelty-induced exploration in trial 1 was defined as the total time spent exploring both of the identical objects. This is a reliable indicator of exploratory behavior. Although this measure is not generally correlated with performance in the testing trials, it is nevertheless an important assessment of general sickness behavior.

The subjects were tested in a longitudinal (i.e., within-group) design at 3 time points (5, 7, and 11 days) after P. berghei ANKA infection or control injection. These time points were chosen on that basis of our previous observations of the time course of the worsening pathological status after infection with the chosen inoculum.

The functional observation battery (i.e., the primary screen) was performed essentially according to the SHIRPA protocol [24, 25] and assessed general arousal, reflexes, and autonomic, motor, and neurological deficits.

Statistical analysis was performed using 1-way repeated-measured analysis of variance (ANOVA), followed by the Fisher protected least significant difference post hoc t test. Post hoc test results are only shown for data in which findings of ANOVA were statistically significant.

Histological analysis and immunostaining

One day after the second and third behavioral tests, infected mice (4 on day 8 and 2 on day 12) were euthanized using CO₂ gas, and their brains were harvested and fixed in 10% buffered formalin. Four control mice were sacrificed on day 12 after injection. Brain specimens of the following widths relative to bregma were obtained: 0.98 to −0.22 mm, −1.82 to −1.94 mm, and −5.88 to −7.2 mm. Specimens were later embedded in paraffin, sectioned into 5-μm slices, and stained with hematoxylineosin.

For quantification of microglial activity, immunohistochemistry analysis was performed using rabbit antibody to ionized calcium-binding adaptor molecule 1 (Iba1) (Wako Chemicals USA), a peptide that is selectively expressed in macrophages/microglia. The sections were deparaffinized and boiled at 95°C for 20 min in sodium citrate solution (DAKO) for antigen retrieval. The sections were then incubated overnight at 4°C with Iba-1 at a dilution of 1:300. A standard avidin-biotin complex method (Vector Immunolab) was used for the secondary antibody (anti-rabbit), using a 1:200 dilution and a 1-h incubation period. Slides were counterstained with hematoxylin after immunolabeling.

To quantify the intensity of Iba1 staining, photographs were taken of the cortex and the tissues surrounding the neuronal cell layers of the hippocampus, using a Nikon Microphot-FXA microscope system and a Nikon digital sight DS-5M camera. The photographs were then analyzed using ImageJ 1.37v (National Institutes of Health) to quantify the staining intensity as a percentage of the region of interest. Five control mice were compared with 7 infected mice, and a t test was used for statistical analysis.

RESULTS

Parasitemia

The average parasitemia (defined as the percentage of parasitized RBCs) in infected mice was 4.12% on day 6 after infection and 9.63% on day 8 after infection.

Visual memory and object recognition

Infected mice began to perform significantly worse than control mice in the object-recognition test on day 7 (figure 1A), based on a repeated-measures ANOVA (F score, 9.9; P < .01). Post hoc tests demonstrated that neither group exhibited significant cognitive impairments on day 5; however, on day 7, there were significant treatment-group effects (P < .01). Visual memory in the infected group was significantly worse than that in the control group (figure 1A).

Figure 1.

A, Infected mice performed worse in the object recognition test of working memory by 7 days after infection (P < .01). Infected animals had significantly worse visual memory than the noninfected control. Infected mice and control mice demonstrated similar levels of novel object exploration (B) and locomotor activity (C).

Most infected mice died by day 11. The remaining 2 infected mice performed worse than control mice on day 12; however, because of the low number of remaining mice, statistical analysis could not be performed. Data from testing on day 11 are included in figure 1A only for the purpose of visual comparison.

Locomotor activity, novel-object exploration, and functional observation battery

To verify the absence of potential confounding effects of infection on nonspecific sickness behavior, several tests were conducted to assess memory performance. The first of these tests assessed locomotor activity in the open field and found no significant difference between infected mice and control mice over time (figure 1C). Second, the total duration of novel-object exploration was assessed during trial 1 of the object-recognition test. Infected and control mice had similar durations of novel-object exploration (figure 1B). Third, the functional observation battery revealed no significant differences in arousal, reflexes, motor coordination, or neurological characteristics (data not shown). These data indicate that the cognitive deficits exhibited by the infected mice were not purely due to nonspecific behavioral effects of sickness caused by the infection.

Histological analysis

Hematoxylin-eosin staining of brain sections obtained from infected mice demonstrated margination of inflammatory cells in the vessels and extravasation of red blood cells in several regions of the brain, including the thalamus, midbrain, and cerebellum (figure 2). Evidence of fornical hemorrhage, which is consistent with damage to the major input/output pathway to the hippocampus, was found in 3 infected mice (75%) sacrificed on day 8 and 2 infected mice (100%) sacrificed on day 12 but was absent in 4 control mice (figure 3).

Figure 2.

A, Hematoxylin-eosin staining of cerebellum of an infected mouse demonstrating dilated vessels with leukocyte margination (arrowheads) and an area of microhemorrhage (arrow). B, Ionized calcium-binding adaptor molecule 1 (Iba-1) immunostaining of the same cerebellar region, demonstrating Iba-1 positive leukocytes (arrows) and activated microglial cells (arrowheads).

Figure 3.

Infected mice (B) demonstrated hemorrhage in the fornix (arrow), which was absent in uninfected mice (A). Sections taken at bregma −1.86 in control mice and Bregma −1.94 in infected mice. A and B show the CA2 region of the hippocampus.

Immunostaining revealed robust Iba-1 reactivity in specimens from infected mice, demonstrating that microglia cell activation was prominent throughout the brain, including the cortex (figure 4B) and the hippocampus (figure 4D). There was extensive microglial activation with retraction and thickening of the microglial processes. In addition, there were Iba-1–positive intravascular inflammatory cells in the cerebellum of an infected mouse, which was consistent with sequestration of monocytes (figure 4). In contrast, Iba-1–positive microglial cells in brains obtained from 5 uninfected mice were highly ramified and uniformly maintained a delicate morphological processes. The percentages (±SD) of the regions of interest in the cortex that stained positive for Iba-1 were 0.8% ± 0.56% in infected mice and 0.23% ± 0.12% in control mice (P < .05); the percentages (± SD) in the hippocampus were 1.07% ± 0.53% in infected mice and 0.21% ± 0.063% in control mice (P < .05).

Figure 4.

Ionized calcium-binding adaptor molecule 1 (Iba-1) staining of the cortex and hippocampus of uninfected control mice and infected mice. In the cortex (A) and hippocampus (C) of control mice, Iba-1–positive microglial cells are small with delicate fine processes (arrows). In the cortex (B) and hippocampus (D) of infected mice, microglial cells undergo morphological changes that reflect activation, such as thickening of the processes and strong Iba-1 reactivity (arrows).

DISCUSSION

Neurological and cognitive impairments are common features of human cerebral malaria, primarily when infection is due to P. falciparum. These deficits may include hemiparesis, ataxia, deep and prolonged coma, intracranial hypertension, extrapyramidal tremor and epilepsy, hearing and visual impairments, and behavioral difficulties and cognitive deficits [2]. Some of the cognitive alterations that persist after the resolution of the infection include impairment in working memory, learning and language deficits, visuospatial and motor deficits, and psychiatric disorders [3, 5, 6, 8–15]. Here, we used the novel object–recognition task, which evaluates working memory, to observe cognitive deficits after infection in an established mouse model of cerebral malaria. In the brains of the infected mice, there were focal areas of inflammation and hemorrhagic damage. This damage, which included inflamed vessels with leukocyte margination and areas of microhemorrhages, was evident in the fornix, thalamus, midbrain, and cerebellum, consistent with the wide-ranging patterns of reported neurocognitive sequelae that ensue in children after malaria [1–6, 8–10, 13].

In addition to the sequestration of inflammatory cells in the brain vasculature of infected mice, activation of microglia in all areas of their brains demonstrated extensive intrinsic brain parenchymal inflammation. Microglial activation, as well as increased inflammatory mediators, can cause neurodegeneration and apoptosis of neuronal cells in the hippocampus, which adversely affects cognitive function [26–28].

Microglia are cells of macrophage/monocyte origin and are important components in the brain’s response to infection. When microglia are activated by different pathogenic stimuli, such as those associated with malaria, they respond in various ways, which include morphological alterations and release of inflammatory mediators [20, 29–31]. This activation corresponds with insults to the brain and can contribute to the encephalopathy of cerebral malaria.

It is difficult to dissect the specific effects of malaria-induced inflammation and damage from the effects of general sickness behavior or peripheral inflammation. It is generally recognized that inflammation resulting from several etiologies can have an adverse effect on cognitive function [32, 33]. In this regard, no currently existing behavior test is unaffected by the effects of systemic inflammatory mediators that result from acute infection or even stress, because cytokines have been shown to interfere with exploratory behavior, locomotor activity, and some hippocampal functions, as well as to induce microglial activation and reactive gliosis [32, 33]. For this reason, our batteries of tests are designed with inherent controls in an effort to minimize the confounding effects of inflammatory mediators and general sickness. Each animal served as its own control for exploratory behavior, locomotor activity, and novel-object preference during the training phase of the trial, when there were no significant differences between infected mice and control mice.

Several brain regions are known to play a role in visual working memory, including the hippocampus [34–37], the posterior parahippocampal region [38], and the perirhinal cortex [39, 40]. In this regard, the hippocampus is an integral part of the brain because it is required for efficient acquisition of visuospatial tasks, as well as for memory encoding and recall [41, 42]. Both functions of the hippocampus are supported by cholinergic and noncholinergic afferents and efferents in the fornix [43, 44]. Indeed, this study detected extensive hemorrhage of bilateral fornices in the infected mice that was absent in the control mice. Lesions of the fornix have been shown to cause marked lasting impairment in both memory and spatial learning; however, fornix lesions are less likely to interfere with short-term recognition memory [45–47].

Although P. berghei ANKA infection is generally considered to be a model of cerebral malaria, there is systemic involvement; thus, it is important to distinguish between nonspecific sickness behavior and specific cognitive deficits. A previous report demonstrated a decrease in locomotor activity in infected mice during the immediate ante mortem period [17], which likely indicated moribund behavior in those mice. However, our observation of the cognitive function of mice at earlier time points after infection did not detect decreased activity in the open field or decreased exploration of novel objects in the training trial. This is consistent with other reports that suggested that locomotor activity is not altered at earlier time points after infection (even at 1.5 days before death) [17]. Furthermore, we demonstrated normal visual placement in the functional observation battery and normal levels of novel-object exploration up to day 11, indicating that novel-object preference, exploratory activity, and visual acuity were intact despite evident cognitive deficits.

The use of the novel object–recognition task has many benefits in this infection model. It does not rely on food or water deprivation to motivate the animal’s behavior, which could be a substantial confounding factor in infected mice. Furthermore, skilled motor coordination, efficient reaction time, and motor stamina are not required, because the tests are rapid and require only normal exploratory behavior. This test is also similar to those conducted to evaluate a range of human disorders [48–50]. Although we did not observe decreased locomotor activity in infected mice, we did not formally test for cerebellar ataxia, a known complication of malaria [1, 4]. Our results clearly demonstrate that the cerebellum is damaged during cerebral malaria. Therefore, more-focused experiments are currently being performed to validate and quantify the associated functional impairment.

Neurocognitive impairment as a result of infection with certain malarial species represents a substantial health, social, and economic burden in countries that are already resource poor, and thousands of children are at risk every year [1–3, 5, 8, 9, 11]. The mechanisms by which the infection causes these deficits are not well understood, although inflammation and vascular impairment have been suggested. Furthermore, there are no uniform standardized methods of identifying children at risk of developing neurocognitive sequelae. Here, we suggest a model of objective cognitive deficit in infected mice, as demonstrated by the decrease in working memory that is associated with brain parenchymal hemorrhage and inflammation, as well as microglial activation and neuronal degeneration. The use of this model will permit the development of strategies to modify neurological outcome that can be transferred to human population-based studies helping to prevent adverse neurocognitive outcomes in those at risk.