NADPH Oxidase: a Possible Therapeutic Target for Cognitive Impairment in Experimental Cerebral Malaria

Long-term cognitive impairment associated with seizure-induced hippocampal damage is the key feature of cerebral malaria (CM) pathogenesis. One-fourth of child survivors of CM suffer from long-lasting neurological deficits and behavioral anomalies. However, mechanisms on hippocampal dysfunction are unclear. In this study, we elucidated whether gp91phox isoform of nicotinamide adenine dinucleotide phosphate oxidase 2 (NOX2) (a potent marker of oxidative stress) mediates hippocampal neuronal abnormalities and cognitive dysfunction in experimental CM (ECM). Mice symptomatic to CM were rescue treated with artemether monotherapy (ARM) and in combination with apocynin (ARM + APO) adjunctive based on scores of Rapid Murine Come behavior Scale (RMCBS). After a 30-day survivability period, we performed Barnes maze, T-maze, and novel object recognition cognitive tests to evaluate working and reference memory in all the experimental groups except CM. Sensorimotor tests were conducted in all the cohorts to assess motor coordination. We performed Golgi-Cox staining to illustrate cornu ammonis-1 (CA1) pyramidal neuronal morphology and study overall hippocampal neuronal density changes. Further, expression of NOX2, NeuN (neuronal marker) in hippocampal CA1 and dentate gyrus was determined using double immunofluorescence experiments in all the experimental groups. Mice administered with ARM monotherapy and APO adjunctive treatment exhibited similar survivability. The latter showed better locomotor and cognitive functions, reduced ROS levels, and hippocampal NOX2 immunoreactivity in ECM. Our results show a substantial increase in hippocampal NeuN immunoreactivity and dendritic arborization in ARM + APO cohorts compared to ARM-treated brain samples. Overall, our study suggests that overexpression of NOX2 could result in loss of hippocampal neuronal density and dendritic spines of CA1 neurons affecting the spatial working and reference memory during ECM. Notably, ARM + APO adjunctive therapy reversed the altered neuronal morphology and oxidative damage in hippocampal neurons restoring long-term cognitive functions after CM.


Introduction
Malaria remains one of the most important infectious diseases till date. About 409,000 deaths were estimated globally in the year 2019, of which 86% cases were accounted from Africa and India (World malaria report 2020) and 1% of them progress to cerebral malaria (CM) [1]. CM is classified under the severe forms of malaria, caused by the infection of Plasmodium falciparum exhibiting fatal complications such as recurrent seizures, delirium, and coma, ultimately leading to death [2]. Cognitive impairment in most survivors after treatment is one of the salient features of CM. Previous reports show that 25% of children survivors exhibit long-term cognitive impairment up to 20-24 months postdischarge [1,3,4]. The mechanism underlying the cognitive decline is not known.
Hippocampus is one of the most complex regions in the brain, which plays an important role in executing learning and memory functions [5]. Cornu ammonis (CA1) is one of the essential regions consisting of pyramidal neurons important for processing hippocampal-dependent memory [5,6].
Dentate gyrus is a critical region of the brain responsible for memory consolidation, formation of new granule neurons for enhanced synaptic plasticity and cognitive functions [7,8]. Neurons consist of dendritic spines, which are synaptic inputs protruding from dendrites that maintain synaptic plasticity, responsible for long-term potentiation (LTP) and long-term depression (LTD) processes essential for cognitive functioning (learning and memory) [9,10]. Short-term or acute stress conditions result in loss of hippocampal CA1 dendritic spine morphology and reduce dendritic spine density resulting in loss of synapses directly correlating with cognitive impairment [11]. Several reports state that CA1 and dentate gyrus regions of hippocampus are highly vulnerable to ischemic neuronal injury at the earliest stage of chronic neurodegenerative diseases such as multiple sclerosis, epilepsy, and Alzheimer's [12][13][14].
Inhibition of NOX has shown an effective response in restoration of neuronal functions in several acute and chronic CNS disorders [30][31][32]. Apocynin (APO) is a phenolic compound isolated from varied plant sources of Apocynum cannabinum and Picrorhiza kurroa. According to Aldieri et al., APO is not a specific inhibitor of NOX but blocks its assembly in membrane exhibiting anti-oxidant and antiinflammatory properties further proven to show neuroprotection in several animal models of neurodegeneration [32][33][34]. Its mode of action involves intracellular decrease in the reduced/oxidized glutathione ratio (GSH/GSSG), used as a potent marker for oxidative stress and preventing activation of nuclear transcription factor NF-κB, an essential mediator of inflammation [35][36][37][38]. Based on the above properties and low toxicity of APO [34], we studied its role in restoration of hippocampal neurons and in preventing cognitive impairment after CM. Artemisinin derivatives are considered the first-line therapy in human cerebral malaria (HCM) [39,40]. Artemether (ARM) is a semi-synthetic derivative of artemisinin extracted from the medicinal herb Artemisia annua. It consists of an endoperoxide bridge responsible for the anti-malarial activity. The heme released during hemoglobin digestion by the parasite interacts with the endoperoxide leading to hemolytic cleavage of peroxide generating free radicals, which cause membrane damage, oxidation of proteins and fats, and inhibition of transcription, replication processes which may likely cause toxicity to host [41,42]. In the present study, we administered APO as an adjunctive to ARM (APO + ARM adjunctive therapy) to counteract the toxicity of ARM and ARM (monotherapy) to two different cohorts of C57BL/6 mice infected with Plasmodium berghei ANKA (PbA), a widely accepted animal model for CM. Further, we conducted cognitive tests to study the sensorimotor functions, working and long-term memory in all the treated animals compared to the control. This study shows the first evidence that APO adjunctive therapy decreased the expression of hippocampal NOX2 with improvement in learning and memory functions after CM.

Animals
A total of 80 C57BL/6 male and female mice of 3-4 weeks old of 15-20 g were purchased from the National Institute of Nutrition (NIN), Hyderabad. All the animals were fed with chow and sterile water ad libitum followed by 12-h light/ dark cycle in the animal house facility at the University of Hyderabad.
(RMCBS), a tool for identifying the subjects of CM and performing rescue treatment. RMCBS experiment consists of 10 behavioral parameters: gait, balance, motor performance, body position, limb strength, touch escape, pinna reflex, toe pinch, aggression, grooming. Each parameter was scored from 0 to 2 depending on the behavior exhibited by an infected mouse considering a total score of 20. Mice with scores ranging from 5 to 12 were considered as symptomatic to CM and subjected to treatment.

Experimental Groups
Animals symptomatic to CM were administered (i.p.) immediately with ARM at a concentration of 25 mg/kg body weight (b.w.) of ARM (dissolved in Arachis oil) and APO at 5 mg/kg (b.w.) (dissolved in 10% of dimethyl sulfoxide (DMSO) and 90% saline) once per day up to 7 days. All the animals that received the adjunctive therapy were transferred to a separate cage labeled as ARM + APO group (n = 23). Another cohort of CM-infected animals (n = 16) was administered i.p. with 25 mg/kg (b.w.) artemether and transferred to a cage labeled as ARM group (n = 27). Normal saline has been widely used as a vehicle for drug administration as it mimics the blood plasma without altering the physiological conditions in animals [43]. Control mice (CON) (n = 15) were administered with saline as vehicle, which was used to dilute iRBCs for infecting C57BL/6 mice. Animals symptomatic to CM were categorized as CM-infected group (n = 11) and were euthanized humanely with ketamine (150 mg/ kg) and xylazine (10 mg/kg) i.p. onset of CM symptoms on days 6-11. CM-infected animals were intracardially perfused with normal saline. Brain samples were stored in 4% paraformaldehyde (PFA) solution for conducting histology studies, some of the brains were kept in Golgi-Cox stain for 17 days at room temperature, and the rest were stored in − 80 °C freezer for performing Western blotting. All the treated animals were subjected to cognitive tests after a survival period of 30 days.

Behavioral Tests
Animals with mild CM symptoms (n = 11) were tested immediately for spontaneous activity (cylinder and adhesive removal test). A battery of cognitive tests were performed to assess learning, memory, and locomotor functions in the treated animals (ARM n = 15; ARM + APO n = 12) in comparison to CON group after survivability of 30 days. All the tests were performed during the light cycle and analyzed by a group of blinded researchers.

Cylinder Test
Each animal from the experimental groups was placed in a clean, transparent plastic open-top cylinder (height: 26 cm; diameter: 16 cm) and recorded video for 3 min. The number of rears (vertical posture of the mouse, standing on its hind limbs) for each mouse was counted. The cylinder was periodically cleaned with 70% alcohol after each trial. The average number of rears and time spent grooming were measured in all the animal groups.

Adhesive Removal Test
This test was performed to study the deficits in sensorimotor response to stimulation adapted for rodents [44][45][46][47]. Small pieces of adhesive labels were stuck beneath the toe of the mouse and placed in the cage. Each animal was subjected to three alternative trials with a 2-min time interval between the tests. If the animal fails to remove adhesive after 120 s, the trial was preceded with the next mouse. The average time to contact the adhesive till its removal from the toe was recorded.

Beam Balance Test
Each animal (except CM group) was guided to walk from one end of the beam to another (40 cm height, 1-m beam, 12 mm width) to study the motor deficits [48] for 2 days and 1 day for testing. Each mouse received three trials on the test day to analyze the rate of "slipping" (any foot coming off the beam) as a motor deficit with a scoring index of "1" (inability to cross the beam), "2" (crossing the beam with dragging limbs), and "7" (crossing the beam with fewer than 2-ft slips). The beam was cleaned with 70% alcohol and removed droppings with clean dry paper.

Barnes Maze
This task is performed to evaluate the long-term spatial memory in rodents [49,50]. This test consists of a circular platform with 20 equidistantly spaced holes along its perimeter (100 cm in diameter). An escape platform was placed under one of the holes leaving the rest empty. Each animal was guided from the maze's center to detect escape platform for 4 min per three sessions up to 4 days (acquisition phase). The maze and the escape platform were cleaned with 70% alcohol following each trial. Animals were subjected to probe trial after removing the escape platform on day 5. The time to detect the escape platform (primary latency) and the number of holes entered before primary latency (primary error) were recorded. The mice were video-recorded and tested individually with the ANY-maze behavioral tracking software version 6.0, Stoelting Co, Wood Dale, USA.

T-Maze Experiment
Mice were subjected to T-maze consisting of three arms measuring diameter 30 × 15 cm height of left-and rightsided goal arms and 40 cm of the start arm. A forced choice of spontaneous alternation was selected where each mouse was gently placed in the start arm for 3 min for habituation [51][52][53]. The mouse was placed in the start arm of the maze after blocking any one side of the arm. The mouse is forced to explore the L-shaped maze for 5 min (acquisition phase). The mouse was placed back in its home cage for 15-min time duration. The maze was cleaned thoroughly with 70% alcohol to remove olfactory cues in the area. During the test phase, the blockage in the arm was removed, and mouse was placed in the start arm and observed for its entry to the arm not visited previously (correct alternation) (test phase). Mouse exploring the arm visited previously during the test phase is considered as wrong alternation. Each mouse was subjected to 6 trials per day for 4 days to study the "correct alternation" and "wrong alternation." The percentage of correct alternation per animal with side preference rate (actively adapt to one side of the arm) was calculated and compared among the groups.

One-Trial Novel Object Recognition Test
On day 1, mouse was placed in the empty square-shaped box made of transparent glass material (dimensions: 30 × 30 × 30 cm) for 20 min (habituation phase). The mouse was removed from the arena and placed back in its home cage. The box was cleaned with 70% alcohol. On day 2, two identical objects were placed 5 cm away from the walls. Mouse was placed in the box for 5 min (familiarization phase). Mouse was placed back in its home cage. The walls of the box along with the identical objects were cleaned thoroughly with 70% alcohol. One of the identical objects was replaced with a novel object having a different shape and color in same position. After 60 min, the same mouse was placed in the center of the arena for 5 min (test phase). The total time spent interacting with both identical objects in the familiarization and novel object in the test phase via sniffing and pawing within a distance of 2 cm was recorded manually [54]. The discrimination index was calculated to estimate the preference towards novel object (novel object interaction/ total interaction with both objects × 100). A value above 50% indicates a greater inclination towards novel object [55].

Histological Staining
After conducting behavioral assays, all the treated animals were euthanized and perfused intracardially with saline and chilled PFA. The whole brain samples of treated groups were collected and stored in 4% PFA and the rest were kept in − 80 °C. Hippocampal sections of 10 µm thickness were subjected to immunofluorescence, Fluoro-Jade C, and hematoxylin and eosin (H&E) staining, while 100 µm sections were used for Golgi-Cox staining. We assessed the changes in the CA1 and dentate gyrus regions of hippocampus in all the experimental groups.

Fluoro-Jade C Staining
Hippocampal brain sections were subjected to the Fluoro-Jade C staining, an anionic dye (AG325 Millipore) which stains only the degenerated neurons [56]. All the sections were immersed in xylene for 45 min and dehydrated in 100% ethyl alcohol with 5% sodium hydroxide for 5 min. All the sections were washed in phosphate buffer saline (PBS) buffer for 2 min and incubated in 0.06% potassium permanganate solution for 10 min with gentle shaking. All the slides were placed in of Fluoro-Jade C stain (FJ-C) solution (0.05% of Fluoro-Jade C powder dissolved in distilled water with 50 µl of acetic acid) for 20 min. All the sections were washed with distilled water for 2 min, dried at 50 °C at dark for 15 min, and mounted with DPX mounting medium. The number of FJ-C-positive neurons was quantified by the cell counter tool of ImageJ software.

Hematoxylin and Eosin Staining
H&E staining is a standard and well-established method for studying morphological changes in tissues during various pathological conditions [57]. According to the protocols mentioned previously, all the brain sections were stained with the Harris hematoxylin stain [58][59][60]. All the images were captured using the Olympus BX-51 microscope at 1000 × magnification.

Golgi-Cox Staining
Brain samples of all the experimental groups after craniotomy (CM samples were collected on days 6-11, and treated samples were collected after performing cognitive tests) were placed in Golgi-Cox stain solution at room temperature for 17 days followed by dehydration in 20-40% sucrose solution. Sucrose dehydrated whole brains impregnated with Golgi-Cox stain were subjected to cryosectioning using Leica Cryomicrotome CM 1850. Hippocampal brain sections were collected at 100 µm thickness for studying the dendritic spine density and 200 µm to analyze the neuronal arborization pattern by Sholl analysis. The cryosections were developed according to the protocol of Zaquot et al. [61]. All the images were captured by the Olympus BX-51 microscope at 1000 × and 400 × magnifications.

Quantification of Dendritic Spine Density of Hippocampal CA1 Pyramidal Neurons
Dendritic spines were quantified from the skeletonized images of Golgi-impregnated CA1 neurons using the cell counter tool of ImageJ software. Spine projections per 10 µm dendrite length of eighteen distal dendrites were considered for quantification from Golgi-impregnated CA1 neuronal images captured at 1000 × magnification using Olympus BX-51 microscope.

Sholl Analysis of Hippocampal CA1 Pyramidal Neurons
Neuronal arborization pattern was quantified by selecting the widely used linear method of the Sholl analysis algorithm [62]. Sholl analysis draws imaginary concentric circles overlapping from soma of the neurons. A highly arborized neuron develops more intersections compared to an altered neuron. The number of intersections (Nm), radius at maximum Nm that is critical radius (r c ), and number of primary dendrites (Np) were calculated using Sholl plugin of ImageJ software. The soma-centered hippocampal pyramidal neurons (n = 16 per group) were converted to 8-bit binary, skeletonized, and subjected to threshold. The image was subjected to Sholl plugin after drawing a line from the soma to the border of the image. Based on the linear Sholl plot, a graph was plotted for average values of Nm on Y-axis against r c on X-axis for each experimental group. Further, we measured the number and total length of apical and basal dendrites of Golgi-impregnated CA1 neurons (n = 16) in all experimental groups by the NeuronJ plugin 1.4.3 tool in ImageJ software.

Evans Blue Staining
Cohorts of ARM (n = 3), ARM + APO (n = 3) on day 30, CM-infected (n = 3) on day 7, and CON (n = 3) mice were injected intravenously (i.v.) with 2 ml/kg of 2% Evans blue dye (SRL Chemicals 46,650). Mice were anesthetized with ketamine (150 mg/kg) and xylazine (10 mg/kg) and perfused intracardially with 1 × PBS buffer after an hour. Brains were surgically removed, weighed, and immersed in 1 ml of formamide solution for 48 h at 37 °C. The samples were homogenized and centrifuged at 14,000 rpm for 20 min at 20 °C. Supernatants were collected carefully and absorbance was measured at 620 nm using a UV-visible spectrophotometer (Hitachi-U2900). The amount of the dye (microgram per milligram of tissue) extravasated was quantified by measuring the absorbance of the dye from the brain samples compared to the standard values of Evans blue dye.

Semi-quantitative Method for Measurement of Gene Expression
Total RNA was isolated from the whole brain samples of all the experimental groups (n = 4 per group) using Trizol reagent (T9424, Sigma-Aldrich) and quantified using Nan-oDropTM 2000 UV-Visible spectrophotometer (Thermo Fisher Scientific). cDNA was synthesized from the 1 µg concentration of RNA using PrimeScript™ 1st strand cDNA Synthesis kit (6110A Takara Bio). cDNA at 0.5 µg/µl (0.5 µl) concentration and 10 pM concentration of forward, reverse primers (1 µl) were added to 5 µl of 2 × Dreamtaq green PCR master mix (K1081 Thermo Fisher Scientific), made up to 10 µl reaction with 3.5 µl of nuclease-free water and subjected to semi-quantitative PCR (Applied Biosystems Veriti 96 well Thermal cycler) at 25 cycles respectively. PCR protocol included denaturation step at 95 °C for 30 s (stage-1), 95 °C for 2 min (stage-2), annealing at respective melting temperature (Tm) for 45 s, and extension at 72 °C for 5 min. Primers were procured from Integrated DNA Technologies (IDT). Primer sequences specific to NOX2 and GAPDH genes are listed in Table 1.

Western Blotting
The whole brain samples were homogenized in sucrose radioimmunoprecipitation assay (RIPA) buffer (

Statistical Analysis
The statistical differences among the experimental groups were calculated by one-way ANOVA (analysis of variance) with post hoc Student-Newman-Keuls test (multiple comparisons) using the GraphPad Prism software version 5.03. The p values less than 0.05 and 0.001 were considered significant.

Rescue Therapy with ARM and ARM + APO Improves Behavioral Patterns After ECM
We observed that the mice infected with PbA showed altered behavior from day 5, such as loss of rearing and exploration (60 s for exploring the corners) with an RMCBS score of 12 ± 1. Later, most of the PbA-infected mice were insensitive to touch escape, pinna reflex, with a sharp decline in the RMCBS score from days 6 to 9 (day 6, 11 ± 0.57; day 7, 6.33 ± 0.88; day 8, 5.3 ± 0.3; day 9, 5 ± 0.57) (Supplementary 1). We administered ARM drug to a cohort of PbAinfected mice at RMCBS score of 14, i.e., from day 6 (ARM group, n = 27) (Supplementary 2). Mice were administered with APO + ARM at a RMCBS score of 14, i.e., from day 6 (n = 23) (Fig. 1a, (Fig. 1b). Mice rescue treated with ARM + APO died of parasite recrudescence on days 10, 11, and 12. Some of the ARM-treated mice died few hours after receiving treatment. Mice symptomatic to CM were moribund between days 6 and 9. Cylinder test and adhesive removal test were included in the battery of cognitive tests as they highly correlate with the dysregulation of nigrostriatal dopaminergic functioning (as our previous study elucidated aberrant dopaminergic receptor signaling in CM) [63], require less training, and can be performed in a single test session [47]. Therefore, we planned to perform these tests on animals with mild symptoms of CM on days 6-8 post-infection. After 30-day survival phase, all the treated and control groups were subjected to beam balance test for training on days 31-33 with 3 trials per day, Barnes maze on days 37-40 as training days with three sessions per day and day 41 as probe trial, T-maze on days 45-47 as training days with 6 trials per day and day 48 as probe trial, and one trial-novel object recognition test with habituation phase on day 52 and familiarization and test phase assessed on day 53. As a general practice, we planned a 3-day interval gap between each test as it is advisable not to conduct subsequent cognitive tests. Sensorimotor tests were conducted between the 3-day gap intervals after performing beam balance test (Fig. 1d).

Apocynin Adjunctive Therapy Improved Cytoarchitecture of Hippocampal Neurons in ECM
The H&E-stained sections of CM-infected group illustrated characteristic neurodegenerative changes such as irregularly shaped or shrunken necrotic CA1 neurons with pyknotic nucleus (Fig. 2a). We observed a prominent increase in neuropil vacuolation in the internal granule layer and hilus of dentate gyrus in CM brain sections collected on days 6-11 post-infection. A reduction in number of pyknotic neurons with vacuolar changes in the neuropil was observed in CA1 and dentate gyrus of ARM + APO group compared to ARM rescue treated group (Fig. 2b). FJ-C staining revealed severe neurodegeneration in the CA1 (38 ± 3.0) (Fig. 3a, c) and the granule layer of the dentate gyrus (30.5 ± 4.5) in CM brain sections. ARM-treated group showed abundant FJ-C-positive neurons in hilus and internal granule layer of the dentate gyrus (26 ± 2) compared to the APO (10.5 ± 1.5) group (Fig. 3b, d). Overall, both the rescue treatments prevented neuronal death after CM ARM + APO treatment (***p < 0.001) showed significant neuroprotection in dentate gyrus and CA1 neurons (*p < 0.05) compared to ARM monotherapy in ECM.

Apocynin Rescue Treatment Improves Hippocampal CA1 Pyramidal Neuronal Morphology and Dendritic Spine Density in ECM
Our results show critical loss of hippocampal CA1 dendritic complexity in CM brain sections collected on days 6-11 post-infection (Fig. 4a). Based on the outcomes of H&E and FJ-C staining, we studied whether ARM and ARM + APO treatment had any positive effect on the dendritic arborization and dendritic spine density of CA1 neurons. Both ARM and ARM + APO therapies improved hippocampal dendritic complexity compared to CM-infected brain sections (Fig. 4a). Sholl analysis studies demonstrated a significant increase in dendritic arborization and primary dendrites of CA1 neurons in ARM + APO (Nm, 14.24 ± 0.60) at r c = 200 µm, Np = 4.42 ± 0.37 compared to ARM (Nm, 11.04 ± 0.53) at r c = 250 µm, Np = 3.98 ± 0.21 and CM (Nm, 4.96 ± 0.08) at r c = 50 µm, Np = 2.67 ± 0.02 (Fig. 4b,d). Golgi-Cox-impregnated CA1 neurons illustrated significant increase in the number and length of basal dendrites in ARM + APO (BsD number, 12.5 ± 1.5; length: 438.0 ± 49 µm) (***p < 0.001) compared to ARM group (BsD number, 4.5 ± 0.5; length: 73.0 ± 16 µm) (Fig. 4e, f) and CM group which exhibited dystrophic neurites (BsD: 3.5 ± 0.5, length: 53.5 ± 7.5 µm) as well Fig. 2 Neuropil deterioration followed by neuroprotection in CA1 and dentate gyrus after APO adjunctive rescue therapy in ECM. a H&E images showing neuropil deterioration with degenerated neurons in CA1 region of hippocampus in CM. Arrowhead showing characteristic neurodegeneration in CA1 region despite ARM rescue therapy. Image representing improved neuropil area of CA1 and reduced neuronal damage after APO adjunctive therapy. b Image representing increased neuropil vacuolation in the internal granule layer and hilus of dentate gyrus in CM. Arrowhead representing characteristic neuronal death in dentate gyrus in ARM rescue treated brain sections and improved neuropil in dentate gyrus in APO adjunctive therapy. Scale bar = 10 µm as length and number of apical (ApD number, 6.5 ± 0.5; length: 188.0 ± 12 µm). There was no significant change in the number and length of apical dendrites of ARM and ARM + APO groups (Fig. 4e, f). Dendritic spine density data indicated a substantial increase in ARM + APO (22 ± 0.34) (***p < 0.001) compared to ARM-treated (12.50 ± 0.5) and CM groups (7.0 ± 0.45) (Fig. 4c, g). Our results show that the rest of the groups had decreased arborization, dendritic spine density, or number and length of apical and basal dendrites as compared to control (dendritic spine density, 35.0 ± 0.31; dendritic arborization, Nm, 16.30 ± 0.9 at r c = 100 µm; Np = 5.02 ± 0.05; apical and basal dendrite length, ApD number: 26.0 ± 2, length: 1915.5 ± 18.5 µm, BsD number: 18.5 ± 1.5, length: 700.5 ± 82.5 µm). Overall, despite both rescue treatments mitigated CM symptoms, Golgi-Cox-impregnated brain sections exhibited a significant difference in dendritic complexity, spine density, and arborization pattern in ARM + APO-treated compared to ARM group.

Apocynin Adjunctive Therapy Reduces NOX2 Expression in CA1 and Dentate Gyrus of Hippocampal Sections in ECM
Double immunofluorescence staining illustrated a significant decrease in the NOX2 fluorescence in dentate gyrus and CA1 in ARM + APO (***p < 0.001, dentate gyrus, 6.48 ± 0.25; CA1, 7.32 ± 0.38) compared to ARM (dentate gyrus, 12.36 ± 0.58; CA1, 16.83 ± 0.45) and CM groups (dentate gyrus, 16.89 ± 1.12; CA1, 25.47 ± 0.35) (Figs. 5 and 6). Mature neurons are identified by NeuN immunostaining for evaluation of its density. NeuN is an epitope of RNA binding fox-1 homolog 3 (Rbfox3), which is exclusively expressed in neurons for its maturation and development [64,65]. Rbfox3 homozygous knockout mice exhibited cognitive deficits and directly correlate with loss of NeuN expression followed by CNS injury [66][67][68][69]. Restoration of NeuN immunoreactivity in dentate gyrus and CA1 regions of hippocampal sections is a clear sign of neuronal recovery in ARM + APO group. We observed a significant loss of NeuN expression which was prominent in hilar regions and dentate gyrus in ARM (***p < 0.001) compared to ARM + APO group (Fig. 5c). Overall, both rescue therapies successfully overcame CM, while ARM + APO treated showed prominent neuroprotection in hippocampal regions compared to ARM group.

Apocynin Rescue Therapy Restores the BBB Integrity and Reduces ROS Levels and Expression of NOX2 in ECM
Loss of blood-brain barrier (BBB) integrity is a salient feature in the pathogenesis of CM [70,71]. Evans blue dye is a well-known vascular marker that does not cross BBB under physiological conditions but binds to albumin and gets across a leaky BBB [72,73]. Brain symptomatic to CM appeared deep blue with increased concentration of Evans blue (9.50 ± 0.5 µg/g brain tissue); at the same time, ARM + APO treated showed a lighter blue coloration with a concentration of 4.5 ± 0.5 µg/g Evans blue dye compared to ARM group (6.1 ± 0.87 µg/g) (Fig. 7a). Overall, our findings represent that ARM + APO-treated group significantly prevented BBB impairment compared to ARM (*p < 0.05) and CM-infected groups (***p < 0.001). We observed lower levels of Evans blue dye in CON group (2 ± 0.03 µg/g).
Recent studies show that deletion of NOX2 gene in animal models shows partial neuroprotection from brain ischemia and traumatic brain injury (TBI) [23,74]. Gene expression studies revealed that both ARM mono (1.20 ± 0.07) and ARM + APO (0.619 ± 0.04) adjunctive therapies were successful in reducing the NOX2 gene expression compared to CM group (***p < 0.001, 2.38 ± 0.27) upon normalization with glyceraldehyde 3-phosphate dehydrogenase (GAPDH) loading control (Fig. 7b). Protein expression data was in line with results of NOX2 gene expression normalized with GAPDH (protein), showing a significant reduction of NOX2 expression in ARM + APO (0.889 ± 0.0089) compared to ARM (*p < 0.05, 1.49 ± 0.09) and CM groups (***p < 0.001, 2.95 ± 0.28) (Fig. 7c). Overall, the gene and protein levels in whole brain lysates showed reduced expression of NOX2 upon treatment with APO adjunctive therapy compared to ARM group in ECM. Previous studies show ROS-induced pathological changes in vascular endothelial damage in CM [75][76][77]. Mean RFU values of DCFDA test data show a significant reduction of ROS in ARM + APO (23,460 ± 2522) (***p < 0.001) compared to ARM (41,465 ± 1663) and CM groups (51,461.5 ± 2538) (Fig. 7d). Overall, based on the DCFDA test, we confirm that APO adjunctive therapy effectively lowers the ROS levels compared to ARM monotherapy in ECM.

Apocynin Rescue Therapy Improves Long-Term Memory and Restores Sensorimotor Coordination After CM
Our analysis from cylinder test exhibited that the number of rearings was significantly increased in both the rescue treated groups (ARM 18 ± 1.15 and ARM + APO 18.33 ± 1.45, ***p < 0.001) (Fig. 8a) compared to CM (1.66 ± 0.88). Furthermore, no change in rate of rearings was observed between ARM-and ARM + APO-treated groups. Adhesive removal test results show a significantly decreased rate of contact 8.5 ± 0.5 and removal time 26.5 ± 1.5 s in ARM + APO (***p < 0.001) compared to ARM treated (contact 39.5 ± 5.5, removal time 85.5 ± 4.5, p < 0.001) and CM (contact 88 ± 2, removal time 90.5 ± 0.5) (Fig. 8b, e). Beam balance test results show that ARM group exhibit a characteristic contralateral foot slipping (***p < 0.001, 7.33 ± 0.66) with increase (*p < 0.05) in the average time (11 ± 1.15 s) (Supplementary 5) to traverse the beam compared to ARM + APO group (slipping rate 3.33 ± 0.3, time to traverse 7 ± 1.15 s) (Supplementary 6, Fig. 8c, f). We observed that CON group occasionally showed slipping and least time to traverse the beam (5.1 ± 0.58 s) compared to rest of the experimental groups. T-maze spontaneous alternation task results show a significant increase in the rate of correct alternation in ARM + APO (60.0 ± 2.88%, ***p < 0.001) group compared to ARM group (43.33 ± 6.0%) (Fig. 8d). We observed that ARM-treated (50 ± 2.88%, ***p < 0.001) group showed a significant strong side preference rate compared to ARM + APO (*p < 0.05) (25.0 ± 2.88%) and APO adjunctive therapy increased NeuN immunostaining with reduction in NOX2 expression after ECM. a Image representing colocalization of NeuN (red) and NOX2 (green) in dentate gyrus of CON brain section. b Loss of NeuN immunoreactivity and increased immunoreactivity of NOX2 in CM-infected brain section. c Enhanced immunoreactivity of NOX2 and loss of NeuN staining in ARMtreated brain section. d Restoration of NeuN immunostaining and reduction of NOX2 levels in APO adjunctive treated brain section. ***p < 0.001. Scale bar = 50 µm CON groups (11.66 ± 1.66) (Fig. 8g). The CON group was used as a reference, resulting in a 70.66 ± 5.81% correct alternation rate compared to the other groups. Our analysis from Barnes maze experiment revealed that ARM + APOtreated group performed primary latency within a shorter time interval (81.76 ± 8.56 s on day 1 and 16.78 ± 7.9 s on day 5) (Supplementary 7) compared to ARM group (127.0 ± 8.74 s on day 1 and 37.32 ± 7.46 s on day 5 during the acquisition phase) (Fig. 8h, i, Supplementary 8). The rate of primary errors was increased in ARM-treated (29.56 ± 7.5, day 1; 27.0 ± 3.4, day 5) compared to the ARM + APO group (19.00 ± 7.2, day 1; 3.45 ± 2.5, day 5) (Fig. 8j, k). The heat maps show that CON (105.0 ± 5.6 s) and ARM + APO groups (64.56 ± 5.3 s) spent more time near the escape platform after primary latency compared to the ARM group (34.5 ± 7.9) during probe trial on day 5. Most of the animals in ARM group exhibited frozen behavior in the center of the maze and detected the escape platform upon several nudges (Fig. 8l). Overall, based on the results of the Barnes maze, ARM + APO group exhibited better spatial reference memory skills with less error rate than the ARM group. Novel object recognition test results revealed that ARM + APO group showed a significant increase in the time spent interacting with the novel object (57 ± 2 s, ***p < 0.001) compared to the ARM group (42.5 ± 2.5 s) during retrieval phase (Fig. 8m). Further, we observed that there was no significant change in the time spent with two identical objects (objects 1 and 2) in all the experimental groups (CON: object 1, 43.5 ± 1.5 s; object 2, 50.5 ± 2.5 s; ARM: object 1, 33.5 ± 1.5 s; object 2, 40.5 ± 2.5 s; ARM + APO: object 1, 36 ± 4 s; object 2, 42.5 ± 2.5 s) during acquisition phase (Fig. 8m). Discrimination index percent data represents that ARM + APO (72%) showed more preference for Fig. 6 APO adjunctive therapy restored NeuN immunostaining in CA1 region after ECM. a Image representing the immunofluorescence staining of NeuN (red) and significant reduction of NOX2 (green) in CA1 region of CON brain section. b Loss of NeuN staining and increased immunoreactivity of NOX2 in CA1 region of CM-infected brain section. c Reduction of NeuN and increased immunoreactivity of NOX2 in ARM-treated section. d Restoration of NeuN and reduced NOX2 immunostaining in APO adjunctive treated brain section. ***p < 0.001. Scale bar = 50 µm recognizing novel object compared to ARM group (57%) (Fig. 8n).

Discussion
Superoxides produced by synaptically localized NOX are vital for hippocampal long-term potentiation (LTP) and hippocampal-dependent memory formation [78,79]. Excess ROS production and chronic inflammatory mechanisms activating microglia fuel the brain's unusual microenvironment, leading to oxidative stress and cognitive decline in several neurodegenerative disorders [80][81][82][83][84]. The pathology of HCM includes sequestration of parasite-infected red blood corpuscles (pRBCs) in the microvessels of the brain, impairing the cerebral microcirculatory blood flow leading to cerebral hypoxia. Persistent hypoxic conditions in the brain affect the brain oxygenation levels associated with profound coma. pRBC sequestration leads to increased synthesis of pro-inflammatory factors, fibrinogen deposits, and activating endothelial cells with the initiation of coagulation cascade breaching blood-brain barrier (BBB) [2,70,[85][86][87]. Animal models of CM do not precisely reproduce the HCM but exhibit some similarities in neuropathological mechanisms involved in CM pathology such as swelling of the brain, hemorrhages due to iRBC sequestration, and neuronal damage in critical regions of the brain [88][89][90]. We assume that extensive hypercoagulation could be an essential factor in this model, potentially leading to the hypoxia and expression of hypoxia-inducible factor-1α (HIF1α), a key regulator of oxygen-regulated gene expression likely to drive NOX2 upregulation in CM [91,92]. NOX2 hypersignaling elevates ROS and pro-inflammatory cytokine levels in activated microglial cells in experimental autoimmune encephalomyelitis (EAE) model of multiple sclerosis [93]. However, previous studies have shown a 6-sevenfold elevation of NOX2 in CA1 region of hippocampus, playing a critical role in the disease progression and altering cognition in neurodegenerative diseases such as cerebral ischemia, Alzheimer's, Parkinson's, Huntington, and amyotrophic lateral Fig. 7 Reduced expression of NOX2 in ECM and restoration of BBB integrity after APO rescue therapy. a Image representing appearance of whole brains stained with Evans blue in all the experimental groups (CM group, day 7; ARM and ARM + APO group, day 30) and corresponding graph showing concentrations of Evans blue dye extravasated. b Image representing NOX2 gene expression and densitometry showing the relative gene levels normalized with GAPDH in the whole brain RNA samples. c Western Blot showing NOX2 protein expression in whole brain lysates with quantification of relative protein levels normalized with GAPDH. d Graph representing quantification of ROS levels by relative fluorescence intensity units in whole brain lysates. *p < 0.05, ***p < 0.001 sclerosis [94][95][96][97][98]. Most of these chronic neurodegenerative diseases involving oxidative stress report involvement of NOX, while till date, no studies have indicated the role of NOX in long-term cognitive impairment in survivors of Alzheimer's, Parkinson's, and Huntington's diseases despite less information [99,100]. Survivors of acute neurodegenerative disorders such as sepsis-associated encephalopathy and neonatal hypoxia show involvement of NOX, experiencing long-term cognitive impairment [101,102]. Earlier studies have shown that free radicals synthesized by NOX2 subunit of NADPH oxidase in macrophages do not suppress parasite progression during malaria pathogenesis and have a minimal role in the gp91l phox−/− mice model [26,[103][104][105]. The current study demonstrates that increased hippocampal NOX2 immunoreactivity is detrimental for hippocampal neurons, negatively affecting learning and memory functions in ECM. However, further research is essential to understand the role of the remaining subunits of NOX in CM pathogenesis.
Dendrites arborize to form stable new synapses maintaining connectivity and are a reliable marker for the integrity of neurons. Extensive dendritic arborization positively correlates with dendritic spine density, which impacts cognitive functions [106,107]. Impaired dendritic arborization leads to compromised neuronal morphology resulting in cognitive impairment [108]. Loss of CA1 and dentate gyrus neurons in the hippocampus is a major cause for learning and memory impairment in several neurodegenerative diseases [5,109,110]. Disruption of brain development during childhood may lead to long-lasting consequences on cognition. Generally, cognitive abilities such as learning, attention, and memory rapidly develop in children within 8-9 years of age [111]. Earlier studies showed that children with a history of CM showed hippocampal injury with poor cognitive outcomes [112][113][114]. A recent MRI report demonstrated bilateral hippocampal sclerosis in HCM with a follow-up of short-term memory loss [115]. However, till date, no effective neuroprotective therapy exists for restoration of cognitive functions after CM.
Existing reports state that NOX subunits alter BBB permeability during neuropathological conditions [116][117][118]. APO treatment significantly reverses BBB permeability in several animal models of brain injury. Interestingly, several ARM adjunctive therapies are proven to improve BBB integrity, reducing disease pathology in experimental models of CM [71,119,120]. Based on the outcome of Evan's blue experiment, we assume that APO could play an essential role in restoring BBB integrity after CM compared to ARM monotherapy. The dosage of 5 mg/kg APO as in the current study has also been studied for its neuroprotective action in various experimental models of stroke and traumatic brain injury [121][122][123][124][125]. APO reduces hippocampal neuronal damage and improves spatial cognition in animal models of traumatic brain injury [125][126][127][128]. Seizures induce oxidative stress mechanisms that disrupt the redox balance in the hippocampus resulting in neurodegeneration. Inhibition of NOX by APO therapy reduced seizure-induced oxidative stress and increased hippocampal neurogenesis [129]. Till date, there is no study conducted on APO as adjunctive in animal models of CM. ARM is also known for its anti-oxidative, anti-inflammatory, neuroprotective properties with few adverse side effects upon treatment [130][131][132]. Earlier studies state that derivatives of artemisinin (artemether or artesunate) administered in mice at a concentration of 25-50 mg/kg exhibited an unusual pattern of neuronal damage such as loss of Nissl substance, swelling of perikaryon, and reduction of normal neurons in the brainstem [133]. Kasaragod et al. have demonstrated that artemisinin derivatives are highly responsible for impairing neuronal activity, reducing the expression of synaptic proteins and cytoskeletal markers in the brain [134]. According to Wang et al., ARM as an adjunct did not help treat psychotic symptoms and cognitive impairment in patients who survived schizophrenia [135]. Similarly, CM-infected patients who survived after artemisinin-based therapy exhibited cognitive impairment [136][137][138]. Based on the outcomes of our study, we understood that 25 mg/kg ARM clears parasite and improves survivability but is ineffective in extending neuroprotection after ECM. Artemisinin derivatives stimulate ROS for mediating cytotoxic action in the parasite [139,140]. Interestingly, Gopalakrishnan et al. proved that NOX plays a potent role in ROS stimulation in the presence of artesunate (one of the artemisinin derivatives) in RBC's infected with Plasmodium species [141]. We assume that ARM might exacerbate the oxygenation conditions in the brain, altering neuroprotective signaling mechanisms during rescue therapy. Nevertheless, further research is necessary to study the role of ARM on NOX regarding the maintenance of physiological processes of neurons in the hippocampal circuit in CM.
Changes in dendritic complexity are salient features reported in several neurodegenerative diseases [142][143][144][145]. According to previous reports, hippocampal neurons are highly vulnerable to oxidative stress, ischemia, and hypoglycemia with characteristic cellular damage [146,147]. Several studies demonstrate that APO acts as a neuroprotective agent inhibiting NOX and microglial activation during pathological conditions [128,129,148]. Lee et al. have shown a significant reduction in hippocampal neuronal death in CA1, CA3, dentate gyrus, and hilus regions by APO treatment in pilocarpine-induced epilepsy models [129]. A recent report shows that APO prevented fibrinogen-induced dendritic spine loss in cortical neurons and dendritic damage in 5XFAD mouse model of Alzheimer's disease [149]. In our study, ARM + APO rescue treatment increased the complexity of hippocampal neuronal density, dendritic spine density of CA1 neurons, and NeuN immunoreactivity after CM. Therefore, we assume that APO adjunctive modulates NeuN expression triggering neuroprotective signaling mechanisms and accelerating a robust increase in hippocampal dendritic complexity after survival from CM.
Aberrant activation of neuronal NOX dysregulate dopaminergic neurotransmission mediated motor coordination in Parkinson's disease [150]. Recently, our group elucidated the mechanism of dysregulation of dopamine receptor signaling and impairment of striatal medium spiny neurons in ECM [63]. According to various clinical trial reports, children who survived CM suffer from motor disabilities after ARM treatment [151][152][153][154]. Several studies have shown that APO treatment prevented motor deficits in animal models of neurodegeneration [150,155,156]. Our results were aligned with the above statement with restoration of locomotor functions after ARM + APO treatment in CM symptomatic animals. However, further research is necessary to elucidate the role of NOX in dysregulation of dopaminergic signaling pathways in the brain responsible for motor coordination in CM pathogenesis. According to Celeste et al., hippocampallesioned rats exhibit a strong side preference with a lower correct alternation rate revealing short-term memory deficits during the T-maze experiment [157]. The discrimination ratio is a valid measure of recognition memory sensitivity for differentiating novel object compared to familiar object for mice subjected to novel object recognition task [158]. The increased primary error rate directly correlates with deficits in spatial learning and memory [50,159]. Hence, we assume that hippocampal neurodegeneration may relate to repetitive side preference patterns that contribute to spatial learning memory loss in the T-maze experiment, increased primary error rate in Barnes maze test, and decreased discrimination index in novel object recognition task in ARM group. Despite similar survival rates, we predict that APO counteracts neurotoxicity caused by the ARM in ECM. However, conducting pharmacodynamic drug-drug interaction studies in larger groups is essential for understanding the pharmacological effect.
Together, our findings lend strong evidence for NOX2 overexpression in hippocampal regions, upon counteracting with ARM + APO rescue treatment, reduced hippocampal NOX2 expression, restored neuronal arborization, and improved cognitive and behavioral functions in ECM. Therefore, we firmly believe APO adjunctive therapy could be a promising therapeutic approach against long-term cognitive impairment after CM.