Complement mediates neuroinflammation and cognitive decline at extended chronic time points after traumatic brain injury

Male C57BL/6 mice (Jackson Labs) underwent controlled cortical impact (CCI) at 11–12 weeks of age, and then subjected to different treatment paradigms. All animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) at the Medical University of South Carolina and Ralph H. Johnson VA Medical Center. Mice were subjected to a 12:12 light: dark cycle and all behavioral testing was performed during the light cycle.

Surgeries were performed as previously described [4, 37]. In brief, coordinates for CCI were set halfway between the bregma and lambda points of the brain and 0.5 mm lateral to the midline in the right hemisphere of the brain. Following craniotomy, contusion was performed with a pneumatic impactor device (Infinite Horizon, Precision Scientific Inc.) with impactor tip size of 3mm. The impact was delivered to the brain on intact dura with parameters as follows: depth = 2.5 mm, velocity = 5.25 m/sec, dwell time of 100 ms, and with a 10° angle. An atlas image of injury location along with photographs of an injured brain at day 180 and a naïve age-matched brain is found in Additional File 1a.

Preparation of the recombinant complement inhibitor CR2Crry has been previously described [6]. All preparations were checked for endotoxin, and complement inhibitory activity confirmed as described [6, 26]. Animals were treated via intraperitoneal (IP) injection at a dose of 10 mg/kg CR2Crry in phosphate-buffered saline (PBS), which was found optimal for the suppression of complement activity in previous acute TBI studies [4]. All treatments started 2 months after the TBI insult and at that time animals were moved from standard size cages to Enriched Environment (EE) cages in order to mimic voluntary motor and cognitive rehabilitation. These cages were double the size of normal housing cages and equipped with ladders, running wheels, plastic tubes, and wooden toys. The setup of an EE cage is shown in Additional File 1b. Groups of animals were subject to different treatment paradigms. One group received a total of 3 doses of CR2Crry every other day (transient group). Another group received continuous treatment, with animals first receiving 3 doses of CR2Crry every other day, followed by a single weekly dose until 6 months after injury (i.e. 4 months after the first administered dose). Vehicle animals received a PBS IP injection in the same schedule as continuous treatment. Each dose of CR2Crry was 10 mg/kg, in accordance with a previous study showing efficacy of a single dose at this concentration when measuring acute outcomes [4], and frequency of dosing was based on our previous dosing schedule in a 2 month paradigm [2].

Barnes maze: This task was used to asses spatial learning and memory as previously described [4, 38]. For analysis, mice were recorded via video camera and data was obtained using the Noldus EthoVision XT system in which total path length was computed. This task was performed on all groups in the last 8 days of the study. Grip Strength: Forelimb grip strength was assessed using a grip strength meter as described [11], with each animal allowed 10 trials, with the mid 6 values of the 10 trials used to compute an average strength (force). This task was performed on vehicle and CR2Crry treated mice 180 days after TBI. Open Field Ambulation: Distance moved was monitored in an open field arena equipped with a Noldus EthoVision XT system that allowed for automated recording, animal tracking, and quantitating behavioral parameters such as distance traveled. Animals were placed in one of the corners of the box and allowed to move freely and explore for 15 min and distance moved was measured. Recordings were acquired 180 days after TBI for vehicle, CR2Crry treated mice, and naïve mice of corresponding age. Gait analysis (Catwalk): Gait function was performed using the automated CatwalkXT system (Noldus Technology Company) as described [13]. Briefly, this system consists of an enclosed walkway on an illuminated glass plate. As the mouse passes from one end of the walkway to the other, scattered light from contact between each paw and the glass is recorded by a high-speed camera set under the glass. Based on the obtained videos of light scattered within a defined area, software computes several gait parameters for each of the hind and front limbs including: print area, base of support, max contact intensity, paw swing, and others for each of three runs, with averaging to obtain values per trial. Recording area was defined with a 10 x 20 cm area and each trial consisted of 3 runs. Run capture duration had a minimum time of 0.5 sec and maximum time of 15 sec, thus any run less than 0.5 sec or greater than 15 sec was excluded. Any run with more than 60% variation was also excluded. 180 days after TBI, recordings were acquired for vehicle and CR2Crry treated mice, and the same task performed for age-matched naïve mice.

Animals were euthanized by isoflurane overdose followed by intracardiac perfusion with cold PBS and subsequently with 4% paraformaldehyde in PBS. Brains were extracted and fixed overnight in 4% paraformaldehyde at 4°C. Brains were then immersed in 30% sucrose in PFA and cut into 40-µm thick coronal sections as previously described [4].

For assessment of lesion volume, 8 serial sections of 40-µm thickness and 700–800 µm apart were selected, mounted on gelatin-coated slides, and following 2 hours of air drying, stained using cresyl violet as previously performed [4, 45]. Tissue sections were imaged at a 4x magnification using Keyence BZ-X710 microscope while applying the tile scan option followed by stitching of all tiles to obtain a full scan of each brain section. To ensure an unbiased approach to quantify lesion, an investigator blinded to treatment groups mapped areas of tissue loss. Lesion volume, ipsilateral tissue remaining volume, and contralateral tissue volume were computed after obtaining area data using NIH Image J from the 8 serial sections. For analysis, percent lesion was computed by dividing the summation of lesion volume from all 8 serial sections by the summation of contralateral tissue volume from all sections multiplied by 2. 3D representative volume filled brain images were prepared from the 8 Nissl-stained sections using the “Free-D” software [5, 10].

IF staining was performed using 40-µm thick slices prepared as above, and sections chosen based on stereological coordinates lying between (+)0.5 mm and (-)2.5 mm relative to bregma. From each animal, 2 full brain slices were selected with matching coordinates [Bregma -1.46 mm (+/- 0.04) and -0.18 mm (+/-0.04)] to stain using free floating technique as previously described [2, 4]. Location of the selected brain sections is shown in a sagittal view of the mouse brain in Additional File 2. The following primary antibodies were used: anti-C3 (Abcam, Cat. #: ab11862, 1:200) (which recognizes intact C3 as well as activated cleaved products including C3b, iC3b, C3d, and C3dg), anti-Iba1 (Invitrogen, Cat. #: PA5-21274, 1:80) (for microglia/macrophage), anti-GFAP (Invitrogen, Cat. #: 13-0300, 1:200) (for astrocytes), and anti-NeuN (Abcam, Cat. #: ab104225, 1:200) (for Neurons). Secondary antibodies used in this study were: donkey anti-rat AlexaFluor 555 nm (Abcam, Cat. #: ab150154, 1:200), donkey anti-goat AlexaFluor 647 nm (Invitrogen, Cat. #: A32849, 1:200), donkey anti-rabbit AlexaFluor 488 nm (Invitrogen, Cat. #: A-21206, 1:200), donkey anti-rat AlexaFluor 488 nm (Invitrogen, Cat. #: A-21208, 1:200), and donkey anti-rabbit AlexaFluor 555 nm (Invitrogen, Cat. #: A-31572, 1:200). Epifluorescence imaging for full scans was performed using 10X magnification on the Keyence BZ-X710 microscope using the tile scan option followed by stitching of all tiles. Quantification of signal was performed using NIH ImageJ by a blinded investigator. Higher magnification images were acquired using a Zeiss LSM 880 confocal microscope using a 40X water objective to capture images at same coordinates with regard to injury site across all samples and different treatment conditions. These images were acquired using the Z-stack feature of the microscope covering 30–40 µm thick regions with a resolution of 1µm per slice within the stack. Images were processed and signal was normalized on the Zen (Zeiss) software and images were analyzed using NIH ImageJ. 3D reconstruction of the 40X imaged fields was than performed on Imaris 9.3 (Bitplane) software. The imaris scrip was used with uniform setting across all slices to obtain surfaces of reconstructed neurons, microglia, and complement C3. The spot analysis was then employed to localize complement deposition fragments and the number of interactions with other surfaces (i.e. microglia and neurons). The number of C3/neuron interactions and the C3/neuron/microglia interactions was then reported.

Statistical analysis and data representation was achieved using the GraphPad Prism 8 (GraphPad, CA) software. Details of statistical tests used for different analyses are described in figure legends. All data in the manuscript are represented as mean ± SEM and P values < 0.05 were considered significant. Power sample size estimation was done as previously and was based on prior studies [2, 4]. This was achieved using G*Power 3.1 [18] and with an acceptable power range of 80–90%. Outliers were defined as having a value greater than the 75 per + 1.5xIQR or lower than 25 per - 1.5xIQR. 75 per corresponds to the 75th percentile of the data, 25 per corresponds to the 25^th percentile of the data, and IQR is the interquartile range which corresponds to the difference between the 75th and 25th percentiles.

To investigate the dynamics of complement activation and its role in promoting chronic neuroinflammation and cognitive decline, we investigated two treatment paradigms of complement inhibition applied in the chronic phase after TBI. In one paradigm, CR2Crry was administered in 3 doses over 1-week starting at 2 months after TBI, and animals monitored until euthanized at 6 months post-TBI (termed 1-week treatment). In the second paradigm, following a 3 dose treatment with CR2Crry as above, treatment was continued on a weekly basis through 6 months after TBI (termed continuous treatment). A vehicle control (PBS) group was also included (refer to schematic, Fig. 1a). After TBI, mice were allowed to recover while being housed in an enriched environment that models cognitive and motor rehabilitation, a frequent clinical scenario (See “Methods” section, Additional File 1b).

Prior studies have shown persistent impairment in spatial learning and memory after TBI [2, 4, 16, 20, 32], and we therefore measured performance of all groups on Barnes Maze task starting at day 172 after TBI and compared performance to age-matched naïve mice (Fig. 1b–d). During the learning phase of the task (first 5 days), mice that received continuous CR2Crry treatment showed a significantly steeper learning slope over time (Path length vs. Time) compared to both vehicle and 1-week treatment groups (Fig. 1b,c, and Additional File 3). There was no difference between vehicle and 1-week treatment groups. Similarly, on retention day testing, mice treated continuously with CR2Crry showed no difference in retention of learned memory compared to naïve, and they performed significantly better than 1-week and vehicle treatment groups (Fig. 1d). Using volume reconstruction of Nissl stained brain sections, we also demonstrated that continuous, but not 1-week treatment, reduced lesion volume measured 6 months after TBI, corresponding to cognitive outcome measures (Fig. 1e,f).

To note, the rationale for using age-matched naïve animals versus sham is that skin incision and craniotomy can lead to proinflammatory, morphological, and behavioral damage compared to naïve animals, and therefore can be considered a component of brain injury [15].

The continuous CR2Crry treatment paradigm that improved cognitive outcome and decreased lesion size was also used to evaluate motor performance after 6 months post injury (with the last 4 months of recovery being in an enriched environment). Control vehicle treatment and naïve age matched mice (9 months) were included in this study. Three behavioral tasks were used to evaluate motor function, namely open field ambulation, forearm grip strength, and gait analysis (Catwalk XT). These tasks cover different domains of motor activity that include fine motor strength, balance, coordination and speed. None of the tests showed any differences between vehicle and CR2Crry treated groups, but some tests did show persistent motor dysfunction at 6 months after TBI compared to naïve controls (Fig. 2).

We next investigated whether the differences between cognitive outcomes in 1-week versus continuous CR2Crry treatment groups correlated with differences in a neuroinflammatory response at 6 months after TBI. We first used high resolution immunofluorescence microscopy to analyze complement (C3) deposition, together with analysis of C3/neuron and C3/neuron/microglia colocalization within the perilesional area (Fig. 3a). C3 deposition was lower in brains from mice treated continuously with CR2Crry compared to vehicle treated mice. Brains from 1-week treated mice also appeared to show decreased C3 deposition, but the difference with vehicle did not reach significance (p = 0.58) (Fig. 3b). Since we previously demonstrated that microglia play a role in phagocytosis and elimination of C3d tagged synapses at earlier time points after TBI [2], we also performed a colocalization analysis. Continuous CR2Crry treatment, but not 1-week treatment, showed a significant decrease in number of C3/NeuN co-localization events, as well as reduced C3/NeuN/Iba1 interactions when compared to vehicle controls (Fig. 3c, d). These data are in support of a similar mechanism of neuron/synapse elimination occurring as long as 6 months after TBI. A 3D representative image of C3/NeuN and C3/NeuN/Iba1 interactions is shown in Fig. 3e. A rotating video of the 3D structures is in Additional File 4.

We analyzed microgliosis at 6 months post TBI by Iba1 staining using unbiased stereology for selection of brain sections from different animals within the area of injury based on location of Bregma as mapped to the Paxino’s brain atlas. Iba1 will not distinguish between microglia and infiltrating macrophages, but the term microgliosis is used here for brevity. Slices from locations corresponding to Bregma − 1.46 mm (± 0.04) (hippocampal plane) and − 0.18 mm (± 0.04) (rostral to hippocampus) are shown (Fig. 4a). Sections more caudal to the hippocampus showed little to no evidence of microgliosis. Compared to vehicle controls, the area of microgliosis was significantly reduced in brains from mice continuously treated with CR2Crry at both stereotactic locations. Brains from 1-week treatment group showed a significant decrease in the area of microgliosis at Bregma − 0.18 mm, but not at Bregma − 1.46 mm (Fig. 4b).

A similar approach was used for analysis of astrogliosis using sections stained for GFAP (Fig. 5a). Compared to vehicle controls, the area of astrocytosis was significantly reduced in brains from mice continuously treated with CR2Crry at the Bregma − 1.46 mm (± 0.04) stereotactic location, and there was a strong trend in reduction at the − 0.18 mm (± 0.04) location (p value = 0.09). In brains from the 1 week treatment group, the difference only reached significance at the Bregma − 1.46 mm (± 0.04) location (Fig. 5b).

In all groups studied, the CCI insult resulted in significant damage to the hippocampus, which plays an important role in cognition. We therefore investigated whether the involvement of the hippocampus in the contralateral hemisphere, which remains structurally intact, may contribute to the observed differences in Barnes maze performance in the different treatment groups. To this end, we analyzed the extent of microgliosis and astrocytosis in the contralateral hippocampus of all treatment groups. At 6 months after TBI, both microgliosis and astrocytosis was significantly increased in the contralateral hippocampus in brains from vehicle treated vs. naïve mice (Fig. 6a, b). Microgliosis, as measured by Iba1 staining intensity, showed a strong trend toward a decrease in the continuously treated group compared to vehicle, but did not reach significance (p value = 0.07). On the other hand, there was a significant decrease in astrocytosis in the contralateral hippocampus from both 1-week and continuous treatment groups. Taken together, these data reveal a potential link between hippocampal microgliosis and astrocytosis and decreased cognitive performance observed in vehicle treated animals, an outcome that can be reversed by complement inhibition.