Antioxidant thioether core-crosslinked nanoparticles prevent the bilateral spread of secondary injury to protect spatial learning and memory in a controlled cortical impact mouse model of traumatic brain injury

Abstract

The secondary phase of traumatic brain injury (TBI) is partly caused by the release of excess reactive oxygen species (ROS) from the primary injury. However, there are currently no therapies that have been shown to reduce the secondary spread of injury beyond the primary insult. Nanoparticles offer the ability to rapidly accumulate and be retained in injured brain for improved target engagement. Here, we utilized systemically administered antioxidant thioether core-cross-linked nanoparticles (NP1) that scavenge and inactivate ROS to reduce this secondary spread of injury in a mild controlled cortical impact (CCI) mouse model of TBI. We found that NP1 treatment protected CCI mice from injury induced learning and memory deficits observed in the Morris water maze (MWM) test at 1-month post-CCI. This protection was likely a result of NP1-mediated reduction in oxidative stress in the ipsilateral hemisphere as determined by immunofluorescence imaging of markers of oxidative stress and the spread of neuroinflammation into the contralateral hippocampus as determined by immunofluorescence imaging of activated microglia and neuron-astrocyte-microglia triad formation. These data suggest NP1-mediated reduction in post-traumatic oxidative stress correlates with the reduction in the spread of injury to the contralateral hippocampus to protect spatial memory and learning in CCI mice. Therefore, these materials may offer an improved treatment strategy to reduce the secondary spread of TBI.

Graphical Abstract

Introduction

Of the numerous biochemical derangements that occur following a traumatic brain injury (TBI), reactive oxygen species (ROS) are one of the most important participants in the complex secondary pathophysiological events that manifest through lipid peroxidation, neuroinflammation, and neurodegeneration [1, 2]. Many treatments have been studied in an effort to protect the surrounding healthy brain from excessive ROS release following TBI; however, there are still no treatment options that have demonstrated an improved outcome in a large, multi-center Phase III trial [3–5]. While there are numerous reasons for the lack of success of TBI clinical trials such as patient variability, biomarker selection, or treatment timing, a major contributor is poor target engagement of delivered therapeutics because of low accumulation and retention in the brain [3–5]. Thus, there is a significant unmet need to develop more effective delivery strategies to overcome the biological barriers that would otherwise inhibit transport of materials into the brain in order to prevent the secondary, long-term damage associated with TBI.

TBI is currently understood in two separate injury phases: primary and secondary. The primary injury occurs directly at the moment of initial impact. An outside force shears axons, disrupts the cell membrane, and causes cell death. The secondary injury, however, is the result of biochemical reactions caused by the primary injury that begins with the release of excitatory amino acids, an influx of calcium ions leading to mitochondrial damage, the release of ROS, and increased expression of cytokines and chemokines, which lead to reactive astrocytes and activated microglia. The excess ROS also causes DNA strand damage and lipid peroxidation, which additionally increase oxidative stress and neuronal cell death within the brain [6]. Secondary injury often results in a positive feedback loop where further neuronal cell death causes increased neuroinflammation leading to additional biochemical derangements and cell death, which can spread to the contralateral hemisphere [7]. Several markers of neuroinflammation are reactive astrocytes, activated microglia, and neuron-astrocyte-microglia triad formation. Neuron-astrocyte-microglia triads are formed when astrocytes form scar tissue around the neurons, and microglia bisect the neurons [8]. This secondary cascade can continue years to decades post-injury [9–12]. Several antioxidant treatments including PEG-conjugated superoxide dismutase, PEG-conjugated catalase (PEG-cat), tirilazad, and their combinations, have shown promise in reducing the spread of secondary injury in pre-clinical studies but have failed to translate into an observed clinical benefit for patients likely because of poor target engagement.

Nanoparticles (NPs) offer one promising approach to overcome the limitations of small molecule drug delivery. Previous studies have shown enhanced permeability and retention of NPs in the brain in mouse models of TBI resulting in improved accumulation and retention of NPs at the site of injury [13–19]. This is thought to be a combined result of the relatively large size of the NP and the severely disrupted blood-brain barrier (BBB) allowing the accumulation of blood components in the brain [20–23]. This large NP size would prevent rapid diffusion out of the brain or removal through glymphatic clearance [24, 25]. Therefore, NPs might improve therapeutic efficacy by increasing target engagement and reducing the spread of secondary injury following TBI. Here, we utilized antioxidant thioether core-cross-linked NPs (NP1), which readily scavenge high levels of ROS [26]. The thioether (i.e., sulfide) functional group readily reacts with hydrogen peroxide and superoxide [27, 28], and thus does not require any additional drug loading, permits scaling up, and is readily modified with other functionality such as for imaging [14]. With the size of NP1 of 16.4 nm and molecular weight of 0.88×10⁶ g/mol [26], we expect the accumulation in the brain similar to previous study by Bharadwaj, et al [16] as well as our previous work with similar NPs [14]. We previously found our mouse models of TBI treated with NP1 showed a reduction in ROS-mediated astrocyte reactivity in vitro, reduction of neuroinflammation in the brain, and improvement in recovery in a mouse model of severe TBI [26]. However, it is still unclear if NP1 treatment reduces the spread of secondary injury in the chronic phase of TBI. We hypothesized that NP1 treatment scavenged the excess ROS released from in the acute phase of TBI, thus reducing secondary TBI markers such as acrolein, the quantity of activated microglia, and neuron-astrocyte-microglia triad formation.

In this study, we utilized a mild controlled cortical impact (CCI) mouse model of TBI [29] to the left cortex to test the ability of NP1 to reduce the spread of secondary TBI markers on the cellular level to the contralateral hippocampus. Since the right hippocampus plays a crucial role in spatial learning and memory [30], we expected improvement in spatial learning and memory if the spread of the secondary injury can be slowed or reduced through NP1 treatment.

Methods and Materials

NP1 Synthesis

Antioxidant thioether core-crossed-linked NPs (NP1) were synthesized as described previously [26]. Briefly, pentaerythritol tetrakis(3-mercaptopropionate) was added to polysorbate 80 and 2,2-dimethoxy-2-phenylacetophenone mixture. The thiol–ene reaction was conducted under UV irradiation to produce PS803SH. The PS803SH was diluted before the addition of pentaerythritol tetraacrylate, hexylamine, and tert-butyl acrylate or hydroxyethyl acrylate to form NP1. The NP1 were purified with Spectra/Por regenerated cellulose dialysis membrane (6–8 kDa cutoff, Fisher Scientific) against acetone, followed by deionized water, before being lyophilized and stored at −20°C until use. NP1 were dissolved to the concentration of 1 mg/mL with sterile Dulbecco’s phosphate-buffered saline (DPBS, Thermo Fisher Scientific Waltham, MA) prior to use. The size of NP1 in PBS pH 7.4 was measured with dynamic light scattering to be 16.4 nm, with molecular weight of 0.88×10⁶ g/mol.

Controlled Cortical Impact Mouse Model of TBI

All animal procedures were performed in accordance with the approval of the University of Nebraska–Lincoln IACUC. Six-week-old male and female C57BL/6J mice (Jackson Laboratory, Bar Harbor, ME, USA) were acclimated for 2 weeks prior to the procedures. Mice were anesthetized with 3% isoflurane gas via inhalation and were maintained at ~1.5% with a nose cone on a stereotaxic frame (David Kopf Instruments, Tujunga, CA, USA). The hair of the scalp was removed with Nair (Church and Dwight Co., Inc., Princeton, NJ, USA), and the scalp was disinfected with a betadine scrub and isopropanol wipes afterward. Lidocaine (0.05 mL of 5 mg/mL) and bupivacaine (0.05 mL of 0.3 mg/mL) were applied to the scalp, and buprenorphine SR (60 μL of 0.5 mg/mL) was given subcutaneously. An approximately 1 cm midline incision was made on the scalp over bregma. An approximately 2 mm craniectomy was made in the skull over the left frontoparietal cortex (2 mm anterior and 2 mm left of lambda) using a surgical drill. A controlled cortical impactor (Hatteras Instruments, Cary, NC, USA) attached to the stereotaxic frame with a 2 mm convex tip was used to impact the brain normal to the dura surface at a depth of 1.5 mm and a velocity of 4 m/s with a dwell time of 80 ms. Any bleeding was controlled and incisions were closed using tissue adhesive. NP1 (100 μL of 1 mg/mL) was injected through the tail vein immediately after the surgery for the NP1 treated group. With the average weight of 22.24 g for male mice and 16.44 g for female mice, the average dose of NP1 administration was 4.5 mg/kg for male mice and 6.1 mg/kg for female mice. The size of each treatment group is as follows: 15 mice in the control group, 21 mice in the untreated CCI group, and 13 mice in the NP1 treated CCI group. This includes both male (8 CCI, 5 NP1, 5 control) and female (13 CCI, 8 NP1, 10 control) in three separate MWM experiments with two experiments consisting of female mice and one of male mice. At the end of the functional outcome studies, brains from mice were collected for histological analyses as described below (Fig. 1A). Weight loss and gain were measured by percentage change for each mouse every day for the first week and then once a week for the next two weeks (Fig. S1).

Figure 1.

Characteristics of NP1, experimental timeline, and unaffected motor coordination learning in CCI and NP1 treatment. A) Putative ROS scavenging reaction of NP1 through the oxidation of thioether bond to a sulfone by ROS. B) Table summarizing the characteristics of NP1 (data from [26]). C) Timeline of behavioral experiments showing time points after CCI when the motor coordination learning (rotoarod) and spatial learning and memory (water maze) were performed as well as when tissues were collected for histology. D) Mice showed no difference in latency to fall regardless of treatment condition but did show an improvement in latency to fall over time (p < 0.01 by two-way ANOVA), indicating motor coordination learning. C) Swim speed measured during water maze performance showed no difference between treatment groups (two-way ANOVA), indicating no difference in the motor coordination between the mice. Data are shown as mean ± SEM.

Rotarod

A Rotor-Rod^™ motor function system (San Diego Instruments, San Diego, CA, USA) was utilized to assess the motor function and learning of the mice prior to all MWM studies. Rotarod trials were started 3 days post-CCI and were repeated daily for 5 days. Mice were placed onto the cylinders, which then began to rotate. The speed linearly increased from 0 to 50 rpm over 5 min. Latency to fall was averaged over 5 separate runs for each animal each day.

Morris Water Maze (MWM)

The MWM behavior analysis was executed based on a previously published protocol about assessing spatial learning and memory [31]. The MWM experiment was started 3 weeks post-CCI, and consisted of two trials: spatial acquisition and reversal. The mice were trained to find the platform using a visible marker before covering the platform with opaque water (white tempura paint) and removing the platform marker. The platform was placed in the southwest quadrant during acquisition trials with the mice starting randomly in the north, east, southeast, and northwest quadrants. The platform was moved to the northeast quadrant during reversal trials with the mice starting randomly in the south, west, northwest, and southeast quadrants. Spatial cues were placed in the north, south, east, and west directions as extra-maze cues. In both acquisition and reversal trials, the mice underwent four trials per day for four days. Male mice (18 total, 8 CCI, 5 NP1, 5 control) and female mice (31 total, 13 CCI, 8 NP1, 10 control) were employed for the MWM trial. Testing for each sex was done separately from each other. Search strategy analysis was done by two researchers separately then results were combined and analyzed as a group. Results were recorded and data analyzed in GraphPad PRISM 7 (GraphPad Software, CA) using the percentages of each experimental group’s usage of each spatial strategy.

Histological Analysis

Mouse brains were collected 32 days post-CCI. The 6 control, 7 untreated CCI, and 8 NP1 treated CCI female mice were deeply anesthetized with 3% isoflurane until there was no reflex movement from a paw pinch. The mice were then transcardially perfused with 4% paraformaldehyde in DPBS. Brain tissue was collected, trimmed to the desired coordinates, and post-fixed in 4% buffered paraformaldehyde for 24 h followed by 30% sucrose in DPBS for 3 days before embedded in optimal cutting temperature (OCT, Fisher Scientific, Waltham, MA) compound, frozen on dry ice, and stored at −80 °C. The brains were sliced coronally at a thickness of 50 μm with a cryotome (Leica Biosystems, Wetzlar, Germany). Sections were washed with DPBS to remove the OCT. The brain slices were blocked with 3% normal donkey serum, 0.3% triton X-100, and 0.1% sodium azide in DPBS. The primary and secondary antibody (Ab) were diluted in the blocking buffer. The brain sections were incubated with primary Ab against NeuN (1:1000, ABN90P, Millipore), GFAP (1:500, ab53554, Abcam), Iba1 (1:1000, 019–19741, Wako), and acrolein (1:500, ab37110, Abcam) for 3 d at 4 °C then washed with blocking buffer 3 times for 5 min each before being incubated with a 1:250 dilution of donkey secondary Ab against goat AF488 (ab150129, Abcam), rabbit AF555 (ab150074, Abcam), and guinea pig AF647 (706605148, Jackson Immunoresearch) for 2 h at room temperature. The brain sections were again washed with the blocking buffer 3 times for 5 min each before being stained with DAPI for 5 min, washed with water, and had ProLong^™ Gold Antifade Mountant (Thermo Fisher Scientific) applied.

Histological quantification

Images were acquired with confocal microscopy (LSM800, Zeiss) at 40x objective lens magnification with z-stack configuration and single stack 20x objective lens magnification for acrolein. Quantitative fluorescence mean intensity analysis of acrolein was performed with ImageJ software on the ipsilateral cortex, CA1, and dentate gyrus (DG). Quantitative image analysis of GFAP+ astrocytes, Iba1+ microglia, and triads were performed on at least two randomly selected viewing fields for each region for each mouse. The quantification of GFAP+ astrocytes, Iba1+ microglia, and triads was counted manually in ImageJ software, and divided by the total area of the image field (mm²) to find the density in the regions. The criteria to characterize an activated microglia cell were defined as cells with modification in cellular structure to be deramified, shortened, twisted, several thickened processes, spheroid shape, and/or an enlarged cell body; as compared to the ramified microglia which have long, thin, and radially projecting processes, as well as small cell body [32–34]. Neuron-astrocyte-microglia triad formations were only counted if astrocytes formed scar tissue around the neurons, and microglia bisect the neurons [8]. The investigator was blinded to the treatment throughout the quantification process. The 3D image was reconstructed with ImageJ software and 3D viewer plugin. Using maximum intensity projections, we first identified regions with overlap between stained neurons, astrocytes, and microglia. We then confirmed the association of the three cell types with 3D stack confocal microscopy to determine if the overlapping signal came from a triad or cells that were on separate planes. Only overlapping signals that were validated as triads were counted.

Dihydroethidium assay

A dihydroethidium (DHE, Thermo Fisher Scientific) assay was performed as previously described with modifications [32] on 2 control, 3 untreated CCI, and 4 NP1 treated CCI female mice. NP1 does not have any inherent fluorescent properties so should not influence the DHE assay. Briefly, DHE was dissolved in dimethyl sulfoxide (DMSO) before further dilution in DPBS. DHE was injected intraperitoneally for a total of 6 μg/g body weight into each mouse at 3 h post-CCI. The mice were perfused at 1 h after DHE injection with ice-cold 4% buffered paraformaldehyde, fixed in 4% buffered paraformaldehyde overnight, immersed in 30% sucrose for 48 h, embedded in OCT compound, and sliced coronally at a 20 μm thickness with cryotome. The brain slices were washed with DPBS three times for 5 min each, stained with DAPI for 5 min, washed with water, and ProLong^™ Gold Antifade Mountant was applied. Images were acquired with confocal microscopy at the excitation wavelength of 488 nm, the emission wavelength of 560–635 nm, and the 10x objective lens magnification. Quantitative fluorescence intensity analysis was performed with ImageJ software on the perilesional and the contralateral hemisphere.

Statistical analysis

All the data in this study were expressed as mean ± standard error of the mean (SEM). A p < 0.05 was considered statistically significant. For latency to escape, Kaplan-Meier survival analysis with Mantel-Cox log-rank test was employed to account for the non-normal distribution of latencies resulting from the 90 second maximum trial duration. Other experiments were evaluated using a two-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test, one-way ANOVA for the probe trials, and multiple student t-tests for DHE and acrolein fluorescence mean intensity between treatment groups. All statistics were analyzed with GraphPad Prism 7 software (GraphPad Software, CA).

Results

Experimental timeline and model validation

A total of 48 mice (18 male and 30 female) were used during the behavioral and motor function experiments separated into three categories: untreated CCI, NP1 treated CCI, and control groups. To determine the effects of NP1 treatment on the chronic secondary injury phase following TBI, we utilized the CCI mouse model of TBI because it is well-characterized in response and recovery and its highly reproducible [35]. Following CCI, mice were immediately injected with 100 μL of 1 mg/mL NP1 (Fig. 1A,B) to ensure high uptake and retention in the injury [13, 14, 26], followed by rotarod analysis during the subacute phase of the injury (4–8 days post-CCI) and MWM during the chronic phase of the injury (18–29 days post-CCI) with perfused brains collected at day 32 (Fig. 1C). Rotarod latency to fall results were used to ensure that no long-lasting motor coordination or balance deficits occurred in these mice that could confound spatial learning and memory analyses using the MWM. Latency to fall was tested during the subacute phase of the injury since the reduction in motor coordination in mice is expected to recover by this time [26]. There was no significant difference between treatment conditions and control mice in rotarod performance (increased latency to fall) during the five test days (Fig. 1D) indicating similar motor coordination and learning as expected with the depth and location of injury, which should not affect the motor cortex or cerebellum. To further assess motor coordination, we measured swim speeds of the mice during their MWM trials and found no significant difference between treatment conditions and control mice (Fig. 1C). Therefore, any changes observed in spatial learning and memory would not be confounded by motor coordination.

NP1 reduce spatial learning deficits in CCI mice

To determine how NP1 treatment affects hippocampal function following a mild CCI, we assessed spatial learning and memory using the MWM during the chronic phase of the injury, where secondary injury would account for observed deficits based on our injury depth. We conducted the MWM study at days 18–22 post-injury for spatial acquisition and days 25–29 post-injury for reversal trials. We employed the log-rank test to account for the non-normal distribution of latencies resulting from the 90 second maximum trial duration [36]. CCI mice took significantly longer to escape on days 3 and 4 of spatial acquisition (days 20 and 21 post-CCI) and day 1 and 3 of the reversal trials (day 25 and 27 post-CCI) as compared to control mice and NP1 treated mice as determined by log-rank test (Fig. 2A; Table S2), indicating impairment in spatial learning and memory. There was no difference in latency to the escape location between conditions during either the acquisition or reversal probe trials (Fig. 2B). From the average of 4 spatial and 4 reversal days, the CCI mice spent 34% of the time in the outer annulus, significantly higher than the control and NP1 treated mice that spent 22% (p < 0.001 by two-way ANOVA) and 24% (p < 0.001 by two-way ANOVA) of the time in the outer annulus respectively (Fig. 2C), suggesting NP1 treatment reduced CCI-induced spatial learning deficits. Additionally, NP1 treated mice spent significantly less time in the outer annulus as compared to untreated CCI mice during the acquisition week probe trials (Fig. 2D). This was further analyzed through search strategy assessment (Fig. 2E) [37]. Control and NP1 treated mice learned to use the spatial search strategies since day 2 and utilized spatial search strategy 56.3% and 54.9% of the time on average throughout the MWM study, whereas CCI mice utilized spatial search strategy, on average, 45% of the time throughout the MWM study, significantly less than control and NP1 treated mice (p < 0.05). This data supports our hypothesis that NP1 treatment protected the contralateral hippocampus from secondary injury during the chronic phase of the injury, allowing mice to utilize hippocampus-dependent spatial search strategies to perform better in the MWM. Analysis of weight gain and loss percentages show a similar increase in percentage of weight gained by the beginning and end of the MWM trials for both sexes (Fig. S1). Sex subanalyses of the behavioral data are presented in Figure S2 with each group being analyzed by sex separately and with the same methods as Figure 2.

Figure 2.

MWM assessment of spatial learning and memory at 1 month post-injury. A) Percent escape during spatial acquisition and reversal trials shown using Kaplan-Meier plots. CCI mice showed significantly higher escape percentage during spatial acquisition days 3 and 4 and reversal days 1 and 3 as determined using the Mantel-Cox method. B) Probe trial latency was not significantly different between any groups using analysis through both two-way and one-way ANOVA. C) Untreated mice showed significant differences in the time spent in the outer annulus throughout both the spatial acquisition and reversal weeks when compared to NP1 treated and control. D) Untreated CCI mice spent more time in the outer annulus during acquisition and reversal trials compared to the control and NP1 treated CCI mice as determined by two-way ANOVA (p < 0.01). E) CCI mice did not rely heavily on spatial search strategies as compared to control and treated mice throughout the MWM study. Search strategies were similar in all groups during reversal trials. * and # indicate a statistical difference as compared to control and NP1, respectively, with one, two, and three symbols indicating p < 0.05, p < 0.01, and p < 0.001, respectively. Data are shown as mean ± SEM.

NP1 reduce markers of post-traumatic oxidative stress in CCI mice

Previous work has shown that mice can effectively perform in the MWM with the function of only the right hippocampus and that bilateral damage of the hippocampus in both hemispheres is needed to achieve a significant reduction in MWM performance [30]. Therefore, our MWM results suggest that untreated CCI mice may have developed contralateral hippocampal damage while NP1 treatment protected against the spread of damage to the contralateral hippocampus. In order to determine the mechanism behind the protection afforded by the NP1, we first determined if the NP1 were able to reduce post-traumatic oxidative stress in the ipsilateral hemisphere. Our previous work demonstrated that NP1 have high antioxidant activity in vitro, with the ROS capacity of 9.93 μmol per mg of NP [26]. Here, we employed a DHE assay to observe the spread of ROS in the acute phase of the injury (4 h post-CCI). We also employed immunofluorescence to observe the presence of acrolein, one of the products of lipid peroxidation, in the chronic phase of the injury (32 days post-CCI). From the normalized DHE fluorescence mean intensity at the perilesional to the contralateral hemisphere, there was a significant increase in the untreated CCI mice compared to the control mice (Fig. 3C). Likewise, there was a significant increase in the fluorescence mean intensity of acrolein in the ipsilateral cortex of untreated CCI mice compared to the control mice, while there was no significant increase between the NP1 treated CCI mice and the control mice (Fig. 3D). We also found that there was a trending increase of the acrolein in the ipsilateral CA1 and little change in the ipsilateral DG of untreated CCI mice compared to the control mice, which might be caused by the depth of the DG away from the lesion as compared to CA1. Thus, this data suggests that NP1 treatment reduced post-traumatic oxidative stress in CCI mice.

Figure 3.

NP1 reduced markers of oxidative stress following CCI. A) DHE staining of the whole brain at 24 h post-injury (IP injection of DHE at 3 h post-injury) shows a marked increase in DHE signal indicating the presence of ROS following CCI, which was reduced in mice that received NP1 treatment immediately following impact (scale bars correspond to 500 μm). B) Acrolein staining in the ipsilateral cortex, CA1, and DG, respectively, reveals lipid peroxidation at 1 month following CCI, which was reduced with NP1 treatment (scale bars correspond to 50 μm). C) The DHE fluorescence mean intensity at the perilesional normalized to the contralateral hemisphere. D-F) The acrolein fluorescence mean intensity quantification in the ipsilateral cortex, CA1, and DG, respectively. Data are shown as mean ± SEM. * indicate a statistical difference as compared to control (p < 0.05), as determined by multiple student t-test.

The density of astrocytes and microglia in CCI mice

The presence of oxidative stress markers often correlates with the reactivity and proliferation of astrocytes and the activation of microglia [38–42]. The density of astrocytes and microglia are elevated in the subacute and the chronic phase of the injury, and thus often used to measure neuroinflammation [43–45]. To determine if NP1 treatment reduced the density of the astrocytes and microglia, we calculated the density of glial fibrillary acidic protein positive (GFAP+; green) and ionized calcium binding adaptor molecule 1 positive (Iba1+; red) cells, the markers of astrocytes and microglia respectively, in the various subfields of the hippocampus in the chronic phase of the injury (32 days post-CCI; Fig. 4). Compared to the control group, the density of astrocytes in the ipsilateral CA2/3 and DG regions of untreated CCI mice was significantly higher (p < 0.01; Fig. 4E), while there was no different in the density of astrocytes in the ipsilateral CA1, contralateral CA1, contralateral CA2/3, and contralateral DG regions of untreated CCI mice (Fig. 4B,E; Table S2). The density of the astrocytes in NP1 treated CCI mice was also similar to the control group in the bilateral CA1, CA2/3, and DG regions (Fig. 4B,E). We found that, compared to the ipsilateral CA1 (240.08 ± 12.4 microglia/mm²), CA2/3 (288.1 ± 17.05 microglia/mm²), and DG (270.49 ± 13.67 microglia/mm²) regions of the control group, the density of microglia in the ipsilateral CA1, CA2/3, and DG regions of the untreated and NP1 treated CCI mice was significantly higher (p < 0.001 and p < 0.01 in both CA1, CA2/3, and DG regions) (Table S3). On the other hand, there was no significant difference in the contralateral CA1, CA2/3, and DG regions between the control, untreated, and NP1 treated CCI mice (Fig. 4C,F; Table S3). This data suggests that NP1 did not significantly reduce the density of astrocyte and microglia in the hippocampus at the chronic phase of the injury compared to the untreated CCI mice.

Figure 4.

GFAP and Iba1 immunostaining and quantification in the hippocampus at 1 month following CCI. A and D) representative maximum projections from confocal image z-stacks of bilateral CA1 and DG of the treatment groups (control, CCI only, and CCI+NP1); magenta is NeuN immunostaining, green is GFAP, and red is Iba1. Scale bar corresponds to 50 μm. B, E, and H) Quantitative analysis of GFAP+ astrocytes (cells/mm²) in bilateral CA1, CA2/3, and DG, respectively. C, F, and I) Quantitative analysis of Iba1+ microglia/macrophage (cells/mm²) in bilateral CA1, CA2/3, and DG, respectively. Data are shown as mean ± SEM. * indicates a statistical difference as compared to control with two and three symbols indicating p < 0.01 and p < 0.001, respectively, as determined by two-way ANOVA.

NP1 reduce the density of activated microglia in CCI mice

To further probe the activation status of the microglia in the chronic phase of the CCI injury (32 days post-CCI), we investigated if NP1 treatment reduced the density of activated microglia in the hippocampus rather than total cell number. Activated microglia change morphology from ramified to amoeboid and play a role in exacerbating the chronic phase of the injury [8, 28, 40, 42, 43]. Therefore, we counted the density of amoeboid microglia in the subfields of the hippocampus (Fig. 5). Compared to the control group, the density of activated microglia in the ipsilateral hippocampal regions of the untreated CCI and NP1 treated CCI mice were significantly higher (p < 0.01 by two-way ANOVA) (Table S4). Meanwhile, the density of activated microglia in the contralateral hippocampus of untreated and NP1-treated CCI mice was not significantly higher, except in the ipsilateral CA1 region of the untreated CCI mice. The density of activated microglia in the untreated CCI mice was significantly higher than the control and NP1 treated CCI mice in the CA1 region (p < 0.01) (Table S4). This data suggests that NP1 mitigate microglial activation in the contralateral hippocampus, especially in the CA1 region, which may play a role in alleviating symptoms in the chronic phase of the injury.

Figure 5.

Quantification of activated microglia at 1 month following CCI in the various subfields of the hippocampus. A–C) Maximum intensity projections from z-stack confocal images of Iba1+ microglia in A) CA1, B) CA2/3, and C) DG. Scale bar corresponds to 50 μm. The insets are the representative image of ramified (inset in the control contralateral CA1 panel) and amoeboid (inset in the CCI contralateral CA1 panel) morphologies of microglia, with scale bar of 10 μm. D–F) Quantification of activated microglia (cells/mm²) as determined by unramified and ameboid morphologies in D) CA1, E) CA2/3, and F) DG. * and # indicate a statistical difference as compared to control and CCI, respectively, with two and three symbols indicating p < 0.01 and p < 0.001, respectively as determined by two-way ANOVA. Data are shown as mean ± SEM.

NP1 reduces neuron-astrocyte-microglia triad formation in CCI mice

To further assess any correlation between the reduction in post-oxidative stress and microglia activation with the improved performance in MWM following NP1 treatment, we investigated if NP1 treatment reduced the markers of neuroinflammation in the bilateral hippocampus through counting the neuron-astrocyte-microglia triads in the chronic phase of the injury (32 days post-CCI) [8, 46–50]. Microglia and astrocytes are responsible for clearing up the dead neurons and thus can be observed as neuron-astrocyte-microglia triads. Compared to the ipsilateral hippocampal regions of the control mice, the untreated CCI mice had significantly higher density of triad formation (p < 0.01), while NP1 treated CCI mice had significantly higher density of triad formation only in the ipsilateral CA1 region (p < 0.01; Fig. 6C; Table S5). For the contralateral hippocampus, the density of triad formation in the contralateral CA1 and DG regions of the untreated CCI mice were significantly higher than that of the control group (p < 0.001; Fig. 6C; Table S5), while there was no difference in the density of triad between control and NP1 treated CCI mice in the contralateral hippocampus. Compared to untreated CCI mice, the density of triad in the ipsilateral CA2/3 and DG regions of NP1 treated CCI mice were 2.01-fold (p < 0.01) and 1.89-fold (p < 0.001) lower, while the density of triad in the contralateral CA1 and DG regions of NP1 treated CCI mice were 2.54-fold (p < 0.01) and 1.81-fold (p < 0.01) lower (Table S5). This data suggests NP1 treatment reduced neuron-astrocyte-microglia triad formation in the contralateral hippocampus of CCI mouse model of TBI compared to the untreated CCI mice in the chronic phase of the injury, which may correlate with less neuroinflammation and better MWM performance in the chronic phase of the injury.

Figure 6.

NP1 reduced neuron-astrocyte-microglia triad formation at 1 month following CCI. A) Representative z-stack confocal images from 50 μm sections. The white circle indicates an identified triad. Scale bar corresponds to 50 μm. B) 3D reconstruction validation of a triad to verify direction interactions of a microglia and an astrocyte with a neuron. C) Quantification of triads throughout various subfields of the hippocampus showing a bilateral increase in triads in CCI mice as compared to controls (p < 0.01 for CA1, DG, and CA2/3 Ipsi), which returns to control levels with NP1 treatment (p > 0.6 except for CA1 ipsilateral where p < 0.05). * and # indicate a statistical difference as compared to control and CCI, respectively, with two and three symbols indicating p < 0.01 and p < 0.001, respectively as determined by two-way ANOVA. Data are shown as mean ± SEM.

Discussion

In this study, we showed that NP1 treatment improved MWM performance, reduced post-traumatic oxidative stress, reduced microglia activation, and reduced neuron-astrocyte-microglia triads in the hippocampus in the chronic phase of the CCI mouse model of TBI. NP1 treatment also reduced ROS in the acute phase of the injury. Our results suggest that NP1 treatment ameliorated the spatial learning and memory of TBI mice when administered intravenously right after the injury.

Previous studies and clinical trials demonstrated the neuroprotection of antioxidant compounds in TBI models where most of the antioxidant compounds are small molecules [4]. Previous data from our group has shown that, although small molecules and NPs are distributed at the same rate in injured brain following intravenous injection, NPs are retained up to three times longer [13, 14]. Therefore, we expect limited target engagement from small molecules due to poor retention. The core of NP1 consists of thioether bonds that are readily oxidized by hydrogen peroxide and superoxide [27, 28], as demonstrated in our previous work [14, 26], as well as by the work of others [51]. Thus, NP1 does not require any additional drug loading and production is easily scaled up.

To validate our CCI mice and ensure any differences we observed in the MWM were likely a result of impaired spatial learning or memory rather than the result of motor deficits during the sub-acute/chronic phase [52–55], we tested mice via rotarod as well as measured their swim speed. We observed no differences in rotarod performance or swim speed in control and CCI mice regardless of treatment condition. The cues around the maze and random starting positions also encouraged the mice to find the island utilizing spatial learning and memory rather than motor learning and memory. Therefore, our MWM results likely reflect differences in spatial learning and memory of the mice due to hippocampal function rather than motor coordination deficits.

Even though every treatment group had the same motor function, untreated CCI mice were less likely to escape the maze within the 90 s time limit (Fig. 2A). Probe trial data indicated no differences in the latency to reach the escape location between treatment groups (Fig. 2B), which suggests minimal deficits in spatial learning and memory in this relatively mild CCI mouse model. Indeed, the CCI parameters we chose to use (2-mm-diameter impact tip with 1.5 mm impact depth at 4 m/s) are more mild than those typically used for a mild CCI procedure (3.4-mm-diameter impact tip with 1.5 mm impact depth at 5.25 m/s) where minimal spatial learning and memory differences are observed compared to uninjured mice [56]. This injury severity was chosen to allow us to better test the ability of NP1 to prevent the spread of secondary injury into contralateral brain. Nevertheless, we were still able to observe spatial learning differences in this model by focusing on the search strategies the mice utilized as well as more powerful statistical comparisons between treatment groups based on the 90 s cutoff times [36]. Untreated CCI mice were more likely to employ non-spatial, rather than spatial, search strategies when searching for the platform (Fig. 2D). This was also confirmed by the fraction of time that the untreated CCI mice spent in the outer annulus of the maze as compared to the other treatment groups (Fig. 2B). This finding is supported by previous work by Brody and Holtzmann, where the CCI mice were more likely to employ non-spatial, rather than spatial, searching strategies compared to control mice [37]. Previous studies also found that a lesion in the right hippocampus reduced spatial learning and memory more than a lesion in the left hippocampus alone, and that bilateral lesioning further reduced spatial learning and memory [30]. Here, the CCI was targeted to the left cortex at a mild depth (1.5 mm) to ensure the right hippocampus was not directly affected by the primary injury. Therefore, our results suggest that the impairment in the performance of CCI mice in the MWM was caused by bilateral hippocampus damage resulting from the spread of secondary injury to the contralateral hippocampus. On the other hand, our results from the NP1 treated CCI mice suggest minimal to no spread of secondary injury to the contralateral hippocampus.

It has been shown that male and female rodents respond differently to TBI, namely in macromolecule and NP retention and accumulation, due to the presence and neuroprotective role hormones like estrogen and progesterone have in TBI [57–59]. Though these findings have been debated [60–62], the possibility for differences in biological sex regarding TBI still requires acknowledgement when testing new therapeutics [58]. Bharadwaj et al. [57], suggests that NP accumulation in male mice is almost 2.5 times smaller than their female counterparts at 24 h post-TBI. In the same study, it was noted that the statistically significant difference between male and female mice was time-dependent and seems to only occur at a time-point of 24 h, with significantly increased accumulation occurring in both sexes at 3 h and 3 days. With the assumption that male mice would have lower NP accumulation 24 h post-TBI, our data suggests a similar effect of NP1 between sexes results in similar outcomes with male mice seeing a more dramatic difference in escape latency and search strategy compared to their untreated peers (Fig. S2), although our study was not designed to compare sex differences and thus was not sufficiently powered to make statistical comparisons here. Therefore, combined with the fact that NP1 was administered within 3 hrs post-CCI where no differences in accumulation between sexes was observed, NP1 is capable of neuroprotection in both male and female mice and any possible reduced accumulation in male mice does not lead to a significantly different outcome.

To clarify the mechanism of improved spatial learning and memory in NP1-treated mice, we assessed two markers of oxidative stress and secondary TBI: ROS and acrolein. The DHE assay is widely used to measure ROS in vivo by fluorescing brighter with higher concentration of superoxide and hydrogen peroxide in the tissue. Thus, we employed a DHE assay to observe the spread and qualitative measurement of ROS in the acute phase of the injury (4 h post-CCI). Acrolein, on the other hand, is the byproduct of oxidation on arachidonic acid and is often used as a marker of lipid peroxidation. We employed immunofluorescence to observe the presence of acrolein in the chronic phase of the injury (32 days post-CCI). ROS is produced in physiological brain activity and ATP production [63], therefore the DHE fluorescence in the control brain can be assumed as the base ROS produced under physiological condition. In order to reduce the variation of the base ROS in individual mice, we normalized the fluorescence from the perilesional to the contralateral hemisphere where we do not expect the spread of secondary injury of TBI in the acute phase. We found that, at 4 h post-CCI, NP1 treatment tended to reduce the level of target ROS in the perilesional area compared to the untreated CCI mice (Fig. 3B), showing strong evidence for high target engagement by NP1. Likewise, NP1 treatment tended to reduce the presence of acrolein in the perilesional area in the chronic phase of the injury compared to the untreated CCI mice (Fig. 3D–F) further supporting high target engagement by NP1 as formation of acrolein and other lipid peroxidation produces is one of the secondary consequences elevated ROS. The reduction of oxidative stress in the acute phase of the injury likely reduced the cascade of post-traumatic oxidative stress. This data, as well as our previous studies using antioxidant NPs [14, 26], suggest that NP1 scavenged ROS and reduced the post-traumatic oxidative stress in CCI mice when administered in the acute phase of the injury.

An elevated density of astrocytes and microglia in the brain has been widely known as one of the markers of neuroinflammation post-TBI, which correlate with increased neuronal cell death, impeded neurogenesis, and exacerbated behavioral deficits following TBI [38–42]. This allows density of astrocytes and microglia to be utilized to determine the scope of neuroinflammation post-TBI. In the event of brain injury, the dead cells from the primary injury release ROS, which signals the activation of microglia. In return, activated microglia release more ROS and inflammatory cytokines to the surrounding cells thereby inducing reactive astrocytes, blood-brain barrier (BBB) permeability, edema, and increased cell death. During normal physiological conditions microglia are mostly present in a resting state. However, even in the case of severe TBI, not every perilesional microglia is activated. Thus, simply calculating the microglia density post-TBI may not be representative of the density of activated microglia. Notably, pro-inflammatory activated microglia are more prevalent than anti-inflammatory activated microglia by three to five times at 1-month post-TBI; thus, an increase in the number of activated microglia is a strong indicator of neuroinflammation [64–66]. Pro-inflammatory activated microglia are also responsible for exacerbating TBI outcomes and promoting the secondary injury cascade. To measure neuroinflammation, we utilized the amoeboid morphology of microglia that represent the pro-inflammatory activated microglia [28, 67–69]. In this study, there was no significant reduction in the density of microglia in the hippocampus by NP1 treatment compared to the untreated CCI mice. However, our results showed that NP1 treatment significantly reduced the density of activated microglia in the contralateral CA1 region of CCI mice (Fig 5). We also found that the density of astrocytes in the contralateral hippocampus of NP1 treated CCI mice were similar to the control mice, while there was a significant increase of astrocytes in the ipsilateral CA2/3 and DG regions of the untreated CCI mice compared to the control mice. Thus, our results suggest that the reduction in post-traumatic oxidative stress led to reduced neuroinflammation in the chronic phase of the injury by NP1 treatment.

Another marker of neuroinflammation is neuron-astrocyte-microglia triad formation [8, 46–50]. Neuron-astrocyte-microglia triad formation has been used as a marker of neuroinflammation associated with aging, lipopolysaccharide exposure, cerebral hypoperfusion, and mouse and rat models of Alzheimer’s disease. Similar to the other triad studies, it is well-known that neuroinflammation and neuronal cell death occur in the CCI mouse model of TBI. Therefore, we counted neuron-astrocyte-microglia triad formation in the hippocampal regions. Neuron-astrocyte-microglia triads are formed when astrocytes form scar tissue around the neurons, and microglia bisect the neurons [8]. Utilizing z-stack stack confocal microscopy, we were able to identify regions where both GFAP and Iba1 positive cells surrounded NeuN positive cells, as shown in Fig. 6B. We found that CCI mice had a significantly higher density of triads than the control group in the hippocampus bilaterally except in the contralateral CA2/3 region. On the other hand, there was no significant difference in the density of triads in NP1 treated CCI mice and the control group except in the ipsilateral CA1 region, which is just below the primary injury location (Table S5). We found that NP1 treatment significantly reduced the density of neuron-astrocyte-microglia triad formation in the contralateral CA1 region, ipsilateral CA2/3 region, and bilateral DG region, while not significantly changing the density of neuron-astrocyte-microglia triad formation in the ipsilateral CA1 region and contralateral CA2/3 region compared to the untreated CCI mice (Table S5). Our results strongly suggest that NP1 treatment reduced neuroinflammation in the chronic phase of TBI in the bilateral hippocampus, which might correlate with the performance improvement of the NP1 treated CCI mice in the MWM study compared to the untreated CCI mice. This correlation is also supported by previous work demonstrating that bilateral lesions caused by lipopolysaccharide injection in the hippocampus significantly reduced the spatial learning and memory in the MWM test [30].

The present study has several limitations. First, the CCI model of TBI is not entirely similar to the most common injury mechanism in humans. Most human TBIs that occur are mild TBIs with no physical penetration of the cranium. However, a new study found that the BBB is breached in the event of repeated mild TBI [70] which might allow NP1 to accumulate in the brain and improve the outcome from the injury [71]. Second, administering treatment right after a TBI may not be possible in human patients when medical support may not be available for hours after injury. However, our recent work revealed that BBB permeability to one type of NP is highest at 3 h post-TBI and the BBB is permeable to NPs up to 48 h post-TBI in this CCI mouse model of TBI [13]. These results suggest that NP1 administration several hours post-TBI might have therapeutic benefits in both experimental and real-world applications. Furthermore, dose-response studies were not performed with NP1, thus the dose of NP1 may be suboptimal to achieve maximal antioxidant properties in the injury where there is typically a bell-shaped response curve for antioxidant therapies [72, 73]. In addition, repeated administrations of NP1 may also increase therapeutic efficacy beyond the single post-CCI administration we used in this study. However, the long retention time of NPs within the injury is likely an advantage of using these NP-based systems although the activity of NP1 over time during its prolonged retention still needs to be determined. Furthermore, any possible toxicity of NP1 will need to be assessed because of these prolonged retention times in the brain. While the toxicity of NP1 has not been tested in humans, we speculate NP1 will show good biocompatibility as it is synthesized from polysorbate 80, which has been used successfully in a wide variety of cosmetic, food, and pharmaceutical applications, although not without some adverse effects [74, 75]. Along with dose-response studies, cell specific accumulation of NP1, such as in microglia or astrocytes, will need to be measured in future studies to help determine how NP1 is reducing oxidative stress in the injury. However, we have previously shown rhodamine B labeled NP1 passively accumulates in the site of injury but without cell specificity [26].

Another limitation was the lack of quantitative measurement of DHE and acrolein experiment beyond the short therapeutic window of antioxidant molecules in TBI treatment post-TBI [76]. Since ROS has a short half-life in biological tissue, delayed treatment of antioxidant molecules or NPs might reduce the neuroprotection efficacy. On the other hand, carbonyl scavenger molecules or NPs might allow for a longer therapeutic window up to the sub-acute and chronic phases of injury given during the acute phase of the injury [77] and should be explored in subsequent work along with additional work comparing sex differences in relation to NP therapeutics.

Conclusion

We found that the impaired spatial learning and memory in the chronic phase of the CCI mouse model of TBI correlated with the spread of secondary injury and neuroinflammation in the hippocampus. We also found that NP1 reduced ROS in the ipsilateral hemisphere at 1-day post-CCI and acrolein in the ipsilateral cortex at 1-month post-CCI. NP1 also significantly reduced the density of activated microglia in the contralateral CA1 region and neuron-astrocyte-microglia triad formation in the contralateral CA1 region, ipsilateral CA2/3 region, and bilateral DG region at 1-month post-CCI in the NP1 treated CCI-mice. This is likely the result of the high ROS scavenging ability of NP1 combined with their high accumulation and long retention time within the injured brain. Therefore, NP1 treatment offers a promising approach to improve outcome following TBI by reducing the spread of secondary injury to the contralateral hemisphere.

Highlights.

• NP1 treatment improves spatial learning and memory in CCI mouse model of TBI.

• NP1 treatment reduces oxidative stress in ipsilateral hippocampus of CCI mice.

• NP1 treatment reduces secondary TBI markers in the hippocampus of CCI mice.

• NP1 treatment reduces neuroinflammation in the hippocampus of CCI mice.