Intravenous Immunomodulatory Nanoparticle Treatment for Traumatic Brain Injury

Abstract

Objective:

There are currently no definitive disease-modifying therapies for traumatic brain injury (TBI). In this study, we present a strong therapeutic candidate for TBI, immunomodulatory nanoparticles (IMPs), which ablate a specific subset of hematogenous monocytes (hMos). We hypothesized that prevention of infiltration of these cells into brain acutely after TBI would attenuate secondary damage and preserve anatomic and neurologic function.

Methods:

IMPs, composed of US Food and Drug Administration–approved 500nm carboxylated-poly(lactic-co-glycolic) acid, were infused intravenously into wild-type C57BL/6 mice following 2 different models of experimental TBI, controlled cortical impact (CCI), and closed head injury (CHI).

Results:

IMP administration resulted in remarkable preservation of both tissue and neurological function in both CCI and CHI TBI models in mice. After acute treatment, there was a reduction in the number of immune cells infiltrating into the brain, mitigation of the inflammatory status of the infiltrating cells, improved electrophysiologic visual function, improved long-term motor behavior, reduced edema formation as assessed by magnetic resonance imaging, and reduced lesion volumes on anatomic examination.

Interpretation:

Our findings suggest that IMPs are a clinically translatable acute intervention for TBI with a well-defined mechanism of action and beneficial anatomic and physiologic preservation and recovery.

Traumatic brain injury (TBI) is a major global and domestic health issue affecting more than 2.5 million people in the United States every year, with more than 5 million Americans currently living with at least 1 TBI-related sequela.^1–3 After the primary injury, there is substantial secondary injury attributable to infiltrating immune cells, cytokine release, reactive oxygen species, excitotoxicity, and other mechanisms.⁴ Despite many preclinical and clinical trials designed to limit such secondary damage, no successful therapies have emerged.^5,6 Here, we introduce immunomodulatory nanoparticles (IMPs) as a strong candidate for a clinically translatable acute pharmacologic intervention for TBI.

IMPs are highly negatively charged, 500nm-diameter particles composed of the Food and Drug Administration (FDA)–approved biodegradable biopolymer carboxylated poly(lactic-co-glycolic) acid (PLGA-COOH). After intravenous administration, IMPs bind to the macrophage receptor with collagenous structure (MARCO) on monocytes, and monocytes bound to IMPs no longer home to sites of inflammation but rather are sequestered in the spleen.⁷ Because hematogenous monocyte-derived macrophages (hMΦs) have been implicated in mediating a major portion of secondary damage after TBI, we examined the effects of intravenous administration of IMPs in models of experimental TBI.^3,8,9

There are numerous different rodent models of TBI that vary in their clinical relevance and scientific reproducibility.¹⁰ We therefore utilized both a more highly reproducible open-skull model, controlled cortical impact (CCI), and a more clinically relevant closed-head weight drop injury (CHI). We found the outcome of IMP infusion to be similar in both models, with both a reduction in the number of infiltrating myeloid cells and reduced proinflammatory polarization of the cells. This resulted in attenuated edema, preservation of brain tissue, and significant preservation of both physiologic visual and motor function.

Materials and Methods

Animal Care and Housing

All animal procedures were performed in accordance with the Public Health Service Policy on Humane Care and Use of Laboratory Animals. All procedures were approved by the Northwestern University Institutional Animal Care and Use Committee. Male 8- to 10-week-old C57BL/6 mice (Charles River Labs, Wilmington, MA) were used and housed at Northwestern University Center for Comparative Medicine in a 12-hour light/dark cycle with a regular chow diet ad libitum.

Controlled Cortical Impact

Mice were anesthetized using 2.5% isoflurane in oxygen. A 3mm-diameter craniotomy was performed 0.5mm rostral and 4mm lateral from the lambda for the visual cortical injury and 4mm lateral from the lambda-bregma midpoint for the sensori-motor cortical injury. Using a stereotaxic frame and a 2mm impactor tip (Impact One stereotaxic impactor; Leica Bio-systems, Buffalo Grove, IL), a 2mm cortical deformation injury was applied (3.00m/s with 30-second dwell time). Buprenorphine analgesic (0.05mg/kg, subcutaneous in 1ml sterile normal saline) was administered daily for 2 days after injury. Enrofloxacin antibiotic (2.5mg/kg, subcutaneous in 1ml sterile normal saline) was administered daily for 3 days after injury to reduce the risk of infection.

Closed Head Injury

Moderate CHI was induced using a modification of a previously published procedure,^11,12 which has been shown to induce persistent gliosis, astrocytosis, and cognitive deficits. Anesthesia was induced using 2.5% isoflurane in oxygen. Mice were placed on prescored aluminum foil stretched across a 38 × 27 × 27cm³ 4-sided acrylic stage immediately underneath an acrylic guide tube. A 240g weight was dropped on the anesthetized mice from a distance of 1.4m, causing the mouse to break through the foil stage and land on sponges placed below. The weight was secured to a string preventing it from bouncing and hitting the mouse a second time. Placement of mice on the foil stage allowed unrestrained acceleration of the head and body on impact. The mice suffered a “glancing” rather than crushing blow from the weight, mitigating confounding negative outcomes seen in other CHI models, such as skull fracture or apnea.¹¹ The righting time was recorded immediately after the injury to ensure that the impact induced the expected brief period of unconsciousness. No CHI animals employed in this study were noted to have skull fracture or large intracranial hematoma on subsequent dissection, and no injury-induced mortality occurred. All injured animals were able to be used for subsequent analysis. Buprenorphine analgesic (0.05mg/kg, subcutaneous in 1ml sterile normal saline) was administered daily for 2 days after injury. Enrofloxacin antibiotic (2.5mg/kg, subcutaneous in 1ml sterile normal saline) was administered daily for 2 days to reduce the risk of infection.

IMP Injection

As previously described by Jeong et al,¹³ PLGA-COOH IMPs (Degradex; Phosphorex, Hopkinton, MA) were diluted in sterile normal isotonic saline to a final concentration of 4.7mg/ml. The PLGA-COOH nanoparticles were used after dilution and were not bound to any drug or chemical. Then 200μl of the dilute IMP was injected via tail vein 2 to 3 hours after TBI. Each IMP group animal received 4 × 10⁹ IMP particles per injection (1.4mg/kg). Additional injections were performed at 24 and 48 hours postinjury (hpi). Control animals received equivalent volume injections of normal isotonic saline at the same time points.

Ladder Rung Walking Test

Mice were habituated with our custom-made ladder rung apparatus for 3 minutes each for the 7 days before injury. For postinjury testing, each mouse performed 3 trials, with 5 minutes of rest, of crossing the ladder rung while being video recorded. Each video was played back in slow motion to count the number of foot slips. A random pattern of ladder rungs was used in each day of testing to ensure the animals were not learning the rung placements.

Visual Evoked Potentials

Anesthesia was induced in an induction chamber and maintained with 2.5% isoflurane gas in oxygen using a nose cone. Mouse body temperature was maintained at 37°C with an isothermal heating pad (Braintree Scientific, Braintree, MA) and monitored with a rectal thermometer (VWR International, Westchester, PA). A 0.2% atropine solution was used to maximally dilate the pupil. A 25-gauge needle with a silver chloride wire electrode was applied subcutaneously in the midline of the occiput, and another electrode was utilized as the reference electrode placed in the snout. Using a dark Faraday cage and covering the contralateral eye with a patch, diffuse light flashes were applied to the ipsilateral eye (33lux, 2Hz) fed through the cage with a fiberoptic cable. The visual evoked potential (VEP) amplitudes were recorded through a low pass filter (1–500Hz with a 60Hz notch filter) and operational amplifier with 10,000 gain (P55 AC preamplifier; Grass-Telefactor, West Warwick, RI) fed through an analog-to-digital converter and analyzed on the workstation (Science 500 Workshop; PASCO Scientific, Roseville, CA). After 10 trials, the opposite eye was stimulated and VEPs recorded.

Flow Cytometry

As previously described, mice were anesthetized with 50mg/kg pentobarbital followed by cardiac perfusion with 30mL of phosphate-buffered saline (PBS).¹³ For CCI, injured section of brain (3mm³ lesional/perilesional area) was microdissected at 72 hours after injury. For CHI experiments, whole brain tissue was processed at 72 hours after injury as well. In both instances, 5 mice per experiment were pooled to perform cell analysis by flow cytometry. Injured sections or whole brains were minced with a razor, pushed through a 100μm filter, and digested at 37°C for 60 minutes in a PBS solution containing 40U/ml of Liberase R1 (Roche Applied Science, Indianapolis, IN) and 50mg/ml DNase I. The resulting cellular suspension was resuspended in 30% Percoll (Sigma-Aldrich, St Louis, MO), overlaid onto 70% Percoll, and centrifuged at 1,000rpm for 25 minutes at 25°C. The cells at the interface were collected, washed, resuspended in fluorescence-activated cell sorting (FACS) buffer (PBS with 2% fetal calf serum), and counted. The number of cell subpopulations in the central nervous system (CNS) were determined by multiplying the percentage of lineage marker–positive cells by the total number of mononuclear cells isolated from the injured brain/section.

To carry out flow cytometry analysis, the Fc receptors were initially blocked using antimouse CD16/32 (0.25μg; eBioscience, San Diego, CA) for 15 minutes at 4°C. Cells were then washed with FACS buffer containing PBS with 1% (volume/volume) fetal bovine serum and 0.1% (weight/volume) NaN₃ (Sigma-Aldrich). Cells were then stained for surface marker for 30 minutes at 4°C using the specified antibodies. These antibodies included CD45, CD11b, CD11c, Ly6C, Ly6G, MHC II, CD86, CD3, CD4, CD8 (all from BioLegend, San Diego, CA, and eBioscience) and MARCO (R&D Systems, Minneapolis, MN). Cells were then washed with PBS, and viability staining was performed using the LIVE/DEAD fixable dead cell stain kit (Invitrogen, Carlsbad, CA). Following viability staining, cells were washed with PBS and were either resuspended in FACS buffer for flow cytometry analysis or were used for intracellular staining to detect inducible nitric oxide synthase (iNOS), eGR2 (both from eBioscience), and CD206 (BioLegend). For intracellular staining, cells were fixed and permeabilized using the FoxP3 staining buffer kit (eBioscience) and then intracellularly stained. Cells were acquired on a BD Canto II and analyzed using BD FACSDiva version 6.1 software. As controls, fluorescence minus one (FMO) was used to place the gates for analysis.

For flow cytometry analysis, cells were first gated according to forward and side scatter and then restricted to singles cells and live cells. Microglia were identified as CD45^int CD11b⁺ Ly6C⁻, infiltrating myeloid cells were identified as CD45^hi CD11b⁺, and infiltrating lymphoid cells as CD45^hi CD11b⁻. On the infiltrating lymphoid population, cells were gated on CD3⁺ CD4⁺ CD8⁻ or CD3⁺ CD8⁺ CD4⁻ to evaluate the different T lymphocyte subpopulations. On the infiltrating myeloid cells, Ly6G⁺ neutrophils were first gated and excluded from the infiltrating myeloid subpopulations. The Ly6G⁻ myeloid cells were divided into CD11c⁺ myeloid dendritic cells (mDCs) and CD11c⁻ monocytes/macrophages. Finally, the monocytes/macrophages were further divided into Ly6C^hi inflammatory monocytes and Ly6c^lo noninflammatory monocytes. From these subpopulations, expression of CD86 and MHC II were evaluated. Expression of iNOS, eGR2, and CD206 were evaluated on the monocytes/macrophages subpopulation.

Lesion Volume Quantification via Cryosectioning

Mice were euthanized via CO₂ inhalation and transcardially per-fused with ice-cold PBS followed by 4% formaldehyde in PBS. Brains were dissected and postfixed for 2 hours in 4% formalde-hyde at 4°C. Samples were dehydrated overnight in 30% sucrose at 4°C and then embedded in optimal cutting temperature matrix (Tissue-Tek, Miami, FL). Starting at 5mm from the rostral pole, 14μm cryostat sections (Leica CM3050S) were taken at 140μm intervals with each succeeding section collected from 1 to 6 slides (Superfrost Plus; Fisher Scientific, Pittsburgh, PA) and repeated until 6 sections filled all 6 slides so that the entire lesion was represented by each slide. Slides number 1 and 4 were imaged (RS-G4; Caliber ID, Rochester, NY), and lesion areas on individual sections were calculated by approximating the native cortical outline and reconstructing and extrapolating volumes from serial sections (opensource software, Fiji Is Just Image J). These areas were then multiplied by the intervening distance between sections to extrapolate the 3-dimensional (3D) lesion volumes.

Immunohistochemistry

As previously described, slides were washed in PBS-Triton (PBS-T) (Triton concentration is given thusly) (0.05% Triton X-100) and incubated with primary antibodies at 4°C overnight in blocking media (5% normal donkey serum in PBS-T). Primary antibodies used were: GFAP (Dako, Carpinteria, CA, Z0334, 1:1,000; Abcam, Cambridge, MA, ab4674, 1:1,000) and MARCO (BAF2956, 1:50).¹³ For GFAP, primary antibodies were visualized with Alexa 647 (far red), 594 (red), and 488 (green) Alexa Fluor conjugated secondary antibodies at 1:1,000 dilution (Life Technologies, Carlsbad, CA). Nuclei were visualized with 4′, 6-diamidino-2-phenylindole (DAPI) at 1:2,000 dilution. Stained sections were mounted in ProLong Gold molecular probes (Invitrogen). Images were acquired with a Leica SP5 AOBS 2-photon equipped confocal scope. Large-format images were stitched using the native Leica Application Suite—Advanced Fluorescence software. For MARCO, biotinylated antigoat secondary was reacted with an avidin-biotinhorseradish peroxidase (Vectastain Elite; Vector Laboratories, Burlingame, CA) and stained with diaminobenzidine. Sections were dehydrated with ethanol, cleared in xylenes, and mounted with Permount (Thermo Fisher Scientific, Waltham, MA).

Immunohistochemistry Image Quantification

Quantification of immunofluorescence images was performed by blinded investigators. Images were quantified using the Fiji open source bioimage software to obtain intensities, astrocyte areas, and cell counts. For MARCO immunohistochemistry (IHC), data were expressed as cells per cubic millimeter. GFAP fluorescence intensity was quantified with consistent laser and acquisition settings between sections and normalized to DAPI staining intensity in the same section. The data are presented are a unitless value reflecting this normalized intensity.

Magnetic Resonance Imaging Acquisition

After induction of anesthesia with 2% isoflurane, each mouse was scanned with 7T Clinscan magnetic resonance imaging (MRI; Bruker BioSpin, Billerica, MA) using a clinical-grade software platform (Syngo; Siemens Medical Solutions, Malvern, PA). Respirations, body temperature, and heart rate were continuously surveilled and checked with an MRI-compatible physiological monitoring system throughout the imaging session with anesthesia continuously delivered through a nose cone. T2-weighted and T2*/R2* imaging sequences were employed with corresponding experiments. For T2*/R2* sequences, a gradient multiecho 3D sequence was used to acquire the maps using the following MR parameters: flip angle = 10; repetition time = 80 milliseconds; multiple values of echo time = 2.7, 6.63, 10.56, 14.49, 18.42, 22.35, 26.28, 30.21, and 31.14 milliseconds. The spatial resolution was chosen to be ~ 180μm isotropic.

MRI Analysis

All analyses were done by blinded investigators using the generated metadata and the freeware image processing software package ITK SNAP (http://www.itksnap.org/). Ventricular volumes were quantified by thresholding the ventricles (CSF) to measure the precise 3D volume of the highlighted regions. To measure the lesion volumes and control for systemic trauma-induced hydrocephalus, the cavitary regions of both sides of the brain were isolated in coronal reconstructions, and the brain matter volume of the injured (left) side was subtracted from the brain matter volume of the uninjured (right) side. To measure the R2* hyperintense regions, the same thresholding parameters were employed on the corresponding sequence files and tabulated with ITK-SNAP. For further 3D reconstruction, opensource MeshLab freeware (http://www.meshlab.net) was employed to differentially color the ventricle and brain matter meshes. Then 3D reconstructed videos were made using screen recording software (FlashBack Express Recorder, https://www.flashbackrecorder.com/express/) while stereotypically manipulating the emerged mesh images.

Statistics

Data were analyzed using GraphPad Prism software (v5.04), and statistical significance was assigned at a predetermined cutoff of p < 0.05. Comparison between any pair of experimental groups was performed using Student t test. Comparisons between 3 or more groups were conducted using one-way analysis of variance (ANOVA) with the Tukey multiple comparison post hoc test. All data are presented as mean standard error of the mean (SEM) unless otherwise noted.

Results

MARCO-Positive Macrophages Infiltrate into the CNS within 72 Hours of TBI

IMPs specifically interact with MARCO⁺ hematogenous monocytes (hMos). An understanding of the temporal kinetics of MARCO⁺ hMos infiltration is necessary to determine appropriate time windows for IMP treatment and to better understand cellular responses to injury in rodent injury models. In injury-naive animals and in mice at 24 and 72 hours after a CCI injury, a microdissected, 3mm³ lesional/perilesional area (or the anatomically similar area from naive animals) was examined by flow cytometry using cell-specific gating strategies. MARCO⁺ cells were first detected in low numbers at 24 hpi and significantly increased at 72 hpi (Fig 1A, B). To further explore the temporal kinetics of MARCO⁺ hMos infiltration, IHC was performed on animals 2, 24, and 72 hours after CCI. No MARCO⁺ cells were noted on IHC at 2 hpi, and at 24 hpi only rare MARCO⁺ cells were seen. However, at 72 hpi a dramatic increase in MARCO⁺ cells was seen within the brain parenchyma adjacent to the injury cavitation site (2 hpi and 72 hpi MARCO stained sections shown in Fig 1C). Rare MARCO⁺ cells were also seen periventricularly, and no MARCO⁺ cells were noted in the contralateral cortex (not shown). Thus, infiltration of MARCO⁺ hMos was minimal within the first 24 hpi and became more apparent at 72 hpi, consistent with prior investigations of monocyte/macrophage response after TBI.¹⁴

FIGURE 1:

Immunomodulatory nanoparticle (IMP) treatment reduces quantity and quality of infiltrating cells into the controlled cortical impact (CCI) lesion. (A) Absolute number of live cell events detected by flow cytometry from a 3mm³ lesional/perilesional area microdissected 24 or 72 hours after CCI injury. A significant increase in MARCO⁺ cells was observed in the central nervous system at 72 hours postinjury (hpi). (B, C) MARCO⁺ cells were identified by immunohistochemistry in the perilesional brain parenchyma at 72 hpi in the CCI model. No MARCO⁺ cells were noted at 2 hpi, and only rare cells were noted at 24 hpi. Cell counts obtained from 3 images immediately adjacent to the injury cavity on each section, 3 sections per animal; n = 3 animals/group. One-way analysis of variance with Tukey post hoc analysis. Asterisk indicates p < 0.05, double asterisk indicates p < 0.01, and triple asterisk indicates p < 0.01. (D–N) The absolute number of live cell events detected by flow cytometry using cell-specific gating from a 3mm³ lesional/perilesional area that was microdissected at 72 hours after CCI injury. (O–S) Further subanalysis of the previously gated cells examined for the percentage of cells expressing MHC II and CD86. (T, U) Pie charts representing all pertinent live cell recorded events compartmentalized by specific cell types and further subanalysis of the populations of myeloid cells in the lesion area in (T) vehicle- and (U) IMP-treated animals. All statistics shown were by 2-tailed t test with α = 0.05. Asterisk indicates p < 0.05, double asterisk indicates p < 0.01, and triple asterisk indicates p < 0.001. Myeloid cells = CD45^hi CD11b⁺; macrophages (Macro)/monocytes (Mono) = CD45^hi CD11b⁺ Ly6G⁻ CD11c⁻; inflammatory monocytes (Inflammatory Mφ) = CD45^hi CD11b⁺ Ly6G⁻ CD11c⁻ Ly6c^hi; noninflammatory monocytes (Noninflammatory φ) = CD45^hi CD11b⁺ Ly6G⁻ CD11c⁻ Ly6c^lo; dendritic cells (mDCs) = CD45^hi CD11b⁺ Ly6G⁻ CD11c⁺; neutrophils = CD45^hi CD11b⁺ Ly6G⁺; lymphocytes = CD45^hi CD11b⁻; and microglia = CD45^int CD11b⁺ Ly6C^lo. Sample size was n = 3 for each group. Each data point (n) is composed of 5 animals’ microdissected lesion areas pooled together.

IMPs Reduce the Number of Infiltrating Immune Cells and Change the Inflammatory Status of the Lesion Site in CCI

To study how IMPs change the immune cell population in the lesion site after injury, IMPs were administered intravenously via lateral tail-vein injection 2 hours after CCI, a clinically relevant time point regarding time to intervention, with additional boluses at 24 and 48 hours after the injury. At 72 hpi, a microdissected 3mm³ lesional/perilesional area was examined by flow cytometry using cell-specific gating strategies. IMP treatment not only reduced the number of cells in the lesion area by 44.0% (see Fig 1), but also drastically changed the number of specific cell types. There was an 84.5% decrease in infiltrating macrophages/monocytes and a 65.8% decrease in overall myeloid lineage cells after IMP treatment. Further, the phenotype of the monocytes that did infiltrate into the lesion site also was altered by the treatment. IMP treatment reduced the number of inflammatory monocytes (determined by the level of Ly6C expression) by 72.4%. A similar reduction in the number of MARCO⁺ cells was noted in IMP-treated animals. Notably, mDCs were significantly reduced, but the number of brain-infiltrating neutrophils was not reduced by IMP treatment. The number of infiltrating lymphocytes was reduced in IMP-treated animals. Most lymphocytes (69–76%) detected in injured animals were CD3⁺ T lymphocytes, which decreased significantly postinjury, including CD4⁺ and CD8⁺ T-cell populations. MHC II expression in monocytes, macrophages, and dendritic cells (mDCs) was markedly reduced in the IMP-treated animals. Thus, in addition to reducing the numbers of myeloid lineage cells, IMP treatment skewed the cells towards a more M2/anti-inflammatory phenotype.^8,15 IMP-treated mice had a corresponding increase in the numbers of myeloid cells in their spleens (not shown), consistent with our previously published data.⁷

IMP Treatment Preserves Functional Tissue in the Visual Cortex after CCI

To determine whether IMP treatment reduced tissue loss after CCI, brains were examined 10 weeks postinjury. Remarkably, IMP treatment resulted in a 44.7% reduction in lesion volume compared to vehicle-treated animals (Fig 2A, B, D). The morphology of subcortical structures such as hippocampus and dorsal striatum were partially disturbed in both groups. However, in vehicle-treated animals, these areas had undergone necrosis with subsequent cavitation, whereas in IMP-treated animals, these areas were less disarranged with a more superficial cavity. Chronic perilesional scarring has been shown to be an important factor in inhibiting axonal regeneration and perturbing native tissue architecture both physically and chemically.^16–18 Because CCI is a focal injury, we also had the opportunity to examine the perilesional glial scar at this 10-week time point. We found that the glial scar was significantly reduced with lower intensity of GFAP staining in IMP-treated compared to vehicle-treated mice at this chronic time point (see Fig 2C, E). Additional animals at an 8-month postinjury time point were examined for evidence of chronic microglial activation via iba1 IHC. No difference in the number of iba1–positive cells in the peri-lesional brain (or anatomically similar area in sham animals) was noted between sham vehicle-treated animals, injured vehicle-treated animals, and injured IMP-treated animals at this chronic time point (data not shown).

FIGURE 2:

Immunomodulatory nanoparticle (IMP) treatment reduces the lesion volume in controlled cortical impact (CCI) injury. (A) Whole brains of vehicle-treated (top row) and IMP-treated (bottom row) animals 10 weeks after visual cortex CCI injury. (B) Representative cryostat brain sections of vehicle- and IMP-treated animals taken through the center of the lesion cavity. The coronal sections correspond to the top middle (vehicle) and bottom middle (IMP) in (A). (C) Representative perilesional confocal images stained for GFAP⁺ astrocytes and 4^′, 6-diamidino-2-phenylindole (DAPI) at 10 weeks postinjury to showcase the attenuated glial scar in the IMP-treated animals. (D) Quantified lesion volumes of the brains in (A) and (B) via cryosectioning and extrapolating to 3 dimensions. (E) Quantification of the GFAP intensity in perilesional areas in 10 weeks postinjury vehicle- and IMP-treated animals. All statistics shown in this figure were by 2-tailed t test with α = 0.05. Asterisk indicates p < 0.05, double asterisk indicates p < 0.01, and triple asterisk indicates p < 0.001. Sample size was n = 7 for lesion volumes and n = 4 for immunohistochemistry.

We next evaluated the function of the cortical tissue preserved by IMP treatment after CCI to the visual cortex using VEPs. We stimulated each eye independently using diffuse flashes and recorded the VEPs from the midline of the occiput. Because rodents do not have substantial binocular representations in their visual cortices, the recorded VEPs were attributable to the visual cortex contralateral to the stimulated eye. We measured the latencies of the P1 and N1 peaks to evaluate the VEP conduction and the area under the curve (AUC) bounded by the waveform to evaluate the extent of potential summation which corresponds to extent of injury (Fig 3).¹⁹ The AUC was greatly decreased in the injured cortex of the vehicle-treated animals compared to the IMP-treated group both during the stimulus and succeeding it. The responses in the cortices of IMP-treated animals did not differ greatly from those in uninjured mice, whereas the responses in vehicle-treated mice were greatly reduced. In contrast, there were no differences between the groups in the voltage summation in the uninjured cortex. Further, the latencies for the P1 and N1 waves were not different among the treatment groups, indicating the integrity of visual pathways rostral to the damaged cortex. Thus, the cortical tissue preserved by IMP treatment retained physiologic function.

FIGURE 3:

Immunomodulatory nanoparticle (IMP) treatment preserves electrophysiologic function in spared visual cortex after parieto-occipital controlled cortical impact injury. (A) Representative trace of a single visual evoked potential (VEP) in an uninjured mouse. (B) A time dilated trace of the VEP to showcase the P1 and N1 waveforms and the bounded area for area under the curve (AUC) measurements. (C, D) Superimposed grand average VEP waveforms of uninjured, vehicle, and IMP groups from the left eye/uninjured right visual cortex (C) and from the right eye/injured left visual cortex (D). (E, F) Time dilated VEPs from (C) and (D), respectively. (G, H) Transformation with the voltage in the uninjured group becoming the x-axis, and the vehicle and IMP waveforms from (E) and (F) expressed in relationship to the uninjured voltage for the left eye (uninjured right visual cortex; G) and the right eye (injured left visual cortex; H). (I) Voltage/potential summated in the time between the P1 response and the end of the stimulus for AUC for the N1 waveform. (J) Voltage/potential summated after the end of the stimulus for AUC for the subsequent minor waveforms. Statistics were by 2-sided t test with α = 0.05. Asterisk indicates p < 0.05 and double asterisk indicates p < 0.01. Sample size was n = 7 in each group.

IMP Treatment Preserves Motor Cortex and Motor Function after CCI

We next examined the effects of IMP treatment after CCI to motor cortex. We used the clinically relevant method of MRI to assess anatomic changes. When imaged at a chronic 16-week time point, the IMP-treated animals had significantly reduced lesion volumes compared to the vehicle-treated animals (Fig 4A–C and Supplementary Video 1–3). Ventricular volumes were larger in both injured groups compared to uninjured animals, but the volumes were significantly smaller in IMP-treated compared to vehicle-treated mice (see Fig 4D). Therefore, although the tissue loss after injury was sufficient to cause hydrocephalus ex vacuo in these animals, the preservation of brain matter after IMP treatment reduced the development of hydrocephalus.

FIGURE 4:

Immunomodulatory nanoparticle (IMP) treatment after unilateral motor cortex controlled cortical impact (CCI) injury decreases lesion volume and attenuates motor deficits. (A) A 16-week postinjury magnetic resonance imaging T2-weighted sequence with serial cuts from the frontal lobe to the caudal aspect of the cerebral cortex to showcase the lesion volume and ventricle size differences among uninjured, vehicle, and IMP groups. (B) Lesion volume quantification of the vehicle and IMP groups. (C) Three-dimensional reconstructions of the brains shown in (A; top) and the ventricles isolated in red (bottom; see Supplementary Videos 1–3). (D) Quantification of the cerebrospinal fluid/ventricular volumes of the different groups. (E) Quantification of ladder rung behavior done at various time points after the CCI measuring percentage of right forelimb (RF) errors. (F) Correlation of lesion volume versus the number of RF strides. (G) Correlation of lesion volume versus the number of RF strides by Digigait. For statistical analysis, (B) uses a 2-sided t test, (D) and (E) use one-way analysis of variance with post hoc Tukey all comparisons, and (F) and (G) use a Pearson correlation. All statistics in this figure used α = 0.05. Asterisk indicates p < 0.05, double asterisk indicates p < 0.01, triple asterisk indicates p < 0.001, and quadruple asterisk indicates p < 0.0001. Sample size in (A) and (D) was n = 4. Sample size in (E) and (F) was n = 5–10 in each group. Blue asterisks indicate comparing uninjured and vehicle groups, red asterisks indicate comparing uninjured and IMP groups, and pink asterisks indicate comparing vehicle and IMP groups; ns = not significant.

To assess motor physiology, we utilized the ladder rung walking test to quantify fine motor behavior in the contralateral forelimb after CCI of the left motor cortex.^20,21 The IMP-treated group had significantly fewer right forelimb (RF) errors than the vehicle-treated group at every time point during 3 weeks of ladder rung testing after the injury and at the chronic 6-month (180 day) time point (see Fig 4). There was a significant correlation between the percentage of RF errors on ladder rung testing and lesion volume. To determine whether IMP treatment resulted in long-term improvements in motor function specific to gait, we performed detailed analysis (DigiGait Imaging System; Mouse Specifics, Boston, MA) of the mice at 16 weeks postinjury.²² A significant correlation was found between the number of RF strides taken and the end lesion volume, again suggesting decreased motor function on the right side at the chronic time point.

Acute Intervention with IMPs Changes the Number and Type of Infiltrating Cells in CHI

Many human TBIs are CHIs. We therefore examined effects of IMP treatment in an additional animal model of TBI, the weight-drop CHI model. CHI leads to less localized damage including contrecoup damage to the contra-lateral hemisphere, and so we performed flow cytometry analysis of cells on whole brain rather than microdissected lesional tissue. Infiltration of MARCO⁺ cells was minimal at 24 hpi in the CHI model (see Fig 5), but the number of MARCO⁺ cells significantly increased at 72 hpi, consistent with prior investigations of monocyte/macrophage response after closed-head TBI.²³ At 72 hpi, the total number of cells did not differ between the IMP and vehicle-treated mice. However, IMP treatment significantly reduced the number of CD45⁺ immune cells by 44.8%. By contrast, there was a significant increase in live CD45^neg cells after IMP treatment by 52.0%. The preparation analyzed by flow cytometry contained both infiltrating cells and resident cells of the CNS, and the increased proportion of live CD45^neg cells suggests that fewer neurons and glia died in the IMP-treated group. Further, IMP treatment significantly reduced the numbers of CD11b myeloid lineage cells by 54.5%, macrophages/monocytes by 56.1% (including MARCO⁺ cells), and lymphocytes by 30.5%. In contrast to the CCI data, only 40% of the lymphocyte population was CD3⁺. All lymphocyte subpopulations, including CD4 and CD8⁺ T-cell populations, were decreased in the whole brain preparation with IMP treatment. IMP treatment also reduced mDCs by 43.4%, which was similar to the findings in the CCI model. IMP treatment also significantly reduced the expression of CD86 on mDCs and MHC II on monocytes/macrophages. MHC II and CD86 are both important mediators of inflammation, thus IMP treatment reduced both the number of infiltrating cells and the inflammatory state of the cells, similar to the findings in the CCI model. Additionally, IMP treatment resulted in an increase in markers associated with M2 macrophages in the CHI model, including CD206 and eGR2, although another marker, iNOS, was not affected.

FIGURE 5:

Mechanism of action of immunomodulatory nanoparticle (IMP) treatment in closed head injury (CHI)/traumatic brain injury (TBI) parallels controlled cortical impact (CCI)/TBI. (A) Absolute number of live cell events detected by flow cytometry from whole brain at 24 and 72 hours postinjury (hpi). After CHI, a significant increase in MARCO⁺ cells was observed in the central nervous system at 72 hpi. (B–O) Total number of live cell events by cell-specific flow sorting from whole brain tissue 72 hpi. (P–T) Subanalysis of the flow sorted cells showing the percentage of MCH II and CD86 expression. (U, V) Pie charts representing all pertinent live cell recorded events compartmentalized by specific cell types and further subanalysis of the populations of myeloid cells in the lesion area in (U) vehicle- and (V) IMP-treated animals. All statistics shown were by 2-sided t test with α = 0.05. Asterisk indicates p < 0.05, double asterisk indicates p < 0.01, triple asterisk indicates p < 0.001. Myeloid cells = CD45^hi CD11b⁺; macrophages (Macro)/monocytes (Mono) = CD45^hi CD11b⁺ Ly6G⁻ CD11c⁻; inflammatory monocytes (Inflammatory Mφ): CD45^hi CD11b⁺ Ly6G⁻ CD11c⁻ Ly6c^hi; noninflammatory monocytes (noninflammatory φ) = CD45^hi CD11b⁺ Ly6G⁻ CD11c⁻ Ly6c^lo; dendritic cells (mDCs) = CD45^hi CD11b⁺ Ly6G⁻ CD11c⁺; neutrophils = CD45^hi CD11b⁺ Ly6G⁺; lymphocytes = CD45^hi CD11b⁻; microglia = CD45^int CD11b⁺ Ly6C^lo. Sample size was n = 3 for each group. Each data point (n) is composed of 5 animals’ microdissected lesion areas pooled together; a total of 15 animals were used for each group. All statistics in this figure used α = 0.05. Asterisk indicates p < 0.05, double asterisk indicates p < 0.01, and triple asterisk indicates p < 0.001.

Some differences emerged between the CHI and CCI models regarding cell infiltration after IMP treatment: (1) there was a significant decrease in the number of neutrophils found in the brain in the CHI experiments with the IMP-treated group (see Fig 5O) compared to CCI experiments in which we observed no difference; (2) there was a reduction of microglia number in IMP-treated mice in the CHI experiments, but there was no overall difference in microglia in the CCI settings; and (3) lymphocyte subpopulations differed between the injury models, with CHI animals having a lower overall proportion of CD3⁺ lymphocytes, and a higher proportion of these CD3⁺ cells were CD8⁺ than in the CCI model. However, in both models, all subpopulations, as well as total lymphocytes, decreased with IMP treatment. Some of these differences in subpopulations may reflect the fact that the entire brain was assessed with the CHI model but only the focally injured tissue was assessed in the CCI animals. Overall, the findings in CHI correlate well with the cellular changes observed following IMP treatment in CCI.

Acute Intervention with IMP Attenuates Edema and Motor Behavior Deficit in CHI

The global injury produced in the CHI model leads to edema and increased intracranial pressure (ICP) in rodents, similar to human injuries.²⁴ We therefore used a quantitative R2* sequencing modality on MRI to evaluate tissue edema at 24 hpi, a time of peak edema. Even with only 1 infusion of IMP/vehicle at 2 hpi, there was attenuated volume of R2* hyperintensity in the IMP group at 24 hours (Fig 6). In CHI, increased intraparenchymal pressure from edema leads to compression of ventricles with a reduction in their size and increased ICP.²⁵ Therefore, as a surrogate for ICP, we used MRI to measure the volumes of the ventricles. The vehicle-treated animals had significantly smaller ventricles than the IMP-treated animals, implying that these animals had increased ICP. Further, a significant inverse correlation was found between the volume of R2* hyperintensity and the ventricular volumes. Finally, because the CHI resulted in damage to both hemispheres, we assessed motor function in all 4 limbs at 24 hpi using the ladder rung assay and found that the IMP group outperformed the vehicle-treated group, suggesting an acutely attenuated motor behavior dysfunction with only 1 IMP infusion.

FIGURE 6:

Immunomodulatory nanoparticle (IMP) treatment acutely attenuates edema and increase in intracranial pressure and preserves motor behavior. (A) Representative slices of quantitative magnetic resonance imaging R2* sequences of vehicle- and IMP-treated animals at 24 hours after closed head injury. (B) Quantification of the total amount of R2* hyperintense volume. (C) Representative 3-dimensional reconstruction of the ventricles of the vehicle- and IMP-treated animals at 24 hours postinjury showing a decreased ventricle size in the vehicle-treated group. (D) Quantification of the ventricular volumes of the vehicle- and IMP-treated group. (E) Pearson correlation of amount of R2* hyperintensity volume versus the ventricular volume. (F) Quantification of the percentage of errors in all limbs at 24 hours postinjury on ladder-rung assay. For statistical analysis, (B) and (D) use 2-sided t tests, (E) uses Pearson correlation, and (F) uses a one-way analysis of variance with post hoc Tukey all comparisons. All statistics in this figure used α = 0.05. Asterisk indicates p < 0.05, double asterisk indicates p < 0.01, and triple asterisk indicates p < 0.001. NS = not significant; n = 5 animals per group.

Discussion

The 2 models employed in this study, CCI and weight-drop CHI, each address unique aspects of human injury. CCI uses a piston to deliver a direct mechanical injury to the exposed dura through a surgical craniotomy,²⁶ providing an excellent model of cortical contusion. CCI reproduces acute aspects of focal brain injury, such as disruption of the blood brain barrier²⁷ and activation of secondary injury processes, such as apoptosis, oxidative stress, and inflammation.²⁸ Chronic pathophysiological consequences, such as ventricular enlargement and overall volume loss, also have been described.²⁹ However, there are limitations to the CCI model. Not all human injury is characterized by contusion, and diffuse effects of CCI are limited. CHI models, on the other hand, model diffuse injury well and replicate associated pathophysiologic processes, including diffuse axonal injury and gliosis.¹² Although extensive focal tissue destruction is not usually seen, diffuse activation of secondary injury processes, including cerebral edema and neuroinflammation, are common in CHI models.

Although sites and distribution differ, neuroinflammation is an important common pathway in the secondary injury cascade across divergent TBI model systems.^8,15,30 Monocyte/macrophage infiltration is a key early step in the development of the inflammatory cascade, with subsequent elaboration of cytokines and infiltration of additional immune components.¹⁴ Our study employed 2 models of brain injury, demonstrating that IMPs are effective across heterogenous mechanisms with differing injury sites and pathophysiology and suggesting that targeted, specific intervention at this upstream point in the neuroinflammatory process has potential for disease modification in TBI.

IMP treatment reduced lesion volume in CCI in both brain sectioning and MRI-based assays. Only symptomatic treatment (decompressive craniotomy, hyperosmolar therapy, temperature control, etc) is currently available for severe TBI; these therapies are incompletely effective at attenuating secondary injury and edema.³¹ No disease-modifying drugs exist for TBI. Broad-spectrum immunosuppressive treatments, such as corticosteroids, produce untargeted suppression of both proinflammatory and anti-inflammatory cells and have failed in clinical trials.¹⁵ Clodronate liposomes, recently evaluated in preclinical studies, are similarly nondiscriminating in ablating monocyte lineage cells and require pretreatment to be effective, obviating clinical translatability.^32–34 IMPs, however, change the overall neuroinflammatory response by altering the number and the proinflammatory phenotype of cells infiltrating the brain, impacting the capacity of these cells to sustain an inflammatory response and direct the recruitment of additional immune cells.

IMP treatment significantly reduces not only the number of monocytes/macrophages but also the number of mDCs and infiltrating leukocytes in the lesion area in both injury models. Because IMPs are specific to MARCO⁺ hMos, the reduction in other infiltrating cells, including lymphocytes and mDCs, is likely due to reduced production of cytokines and chemokines by hMos.^8,15 IMP treatment also significantly altered the inflammatory status of infiltrating cells toward a less inflammatory profile, trending to a more M2-like phenotype. Prior studies have suggested that microglia-derived macrophages assume a more M2-like phenotype and mediate anti-inflammatory and supportive functions following TBI.^35–37 Our data also agree with prior findings that hMos ablation leads to an attenuated glial scar chronically, suggesting a protective tissue response.^38,39 In the CHI model, there was a reduction in the number of neutrophils and microglia in the brain that was not observed in the CCI model. This finding may be because CHI promotes more systemic inflammation, causing a greater number of cells to acutely exit the bone marrow, amplify the relative change in the number of later infiltrating neutrophils, and change the neuroinflammatory milieu enough so that microglia do not proliferate following CHI.

Importantly, we show electrophysiologic evidence of preserved visual function via VEPs in mice injured in the visual cortex. The greater amount of voltage/potential summation in the injured visual cortex in the IMP-treated group in parallel with the reduction in lesion volume strongly indicates that the preserved tissue retained some native physiologic cortical processing capabilities similar to occipital injuries in humans.¹⁹ VEPs in the uninjured cortex did not differ among the groups, indicating the specificity of the changes to the anatomic injury. Similarly, the latencies of the P1 and N1 waves were not significantly altered, suggesting that the more rostral visual conduction system was not significantly perturbed, supporting the relationship of the changes in VEPs to the focal injury. We did not see a difference in fractional anisotropy via diffusion tensor imaging sequences in our MRI studies of CCI-injured animals at 16 weeks postinjury (data not shown), further suggesting that the white matter tracts were similar between the treatment groups. Interestingly, VEPs in IMP-treated animals were so well preserved that they were similar to control values, despite the presence of a smaller but still significant lesion cavity. It is possible that flash VEP recorded from a surface electrode is insufficiently sensitive to smaller injuries and that additional electrophysiologic testing, such as pattern VEP or cortical electrodes, might detect a residual difference between the groups. Regardless, these results demonstrate significant preservation of function after IMP treatment.

We used clinically relevant MRI techniques to show the reduction in the lesion volume and significant correlations between the end lesion volume and the motor performance acutely, subacutely, and chronically. MRI also allowed us to investigate cerebral edema in the CHI model. TBI-induced cerebral edema is clinically prevalent and causally related to poor outcomes. IMP treatment resulted in a significant reduction in both measures of edema and imaging surrogates for ICP. These measures significantly correlated with each other as well as with significant global motor function preservation. We believe that the innovative use of such multiple and clinically relevant outputs should be considered for future studies of experimental CHI.

Our anatomic, electrophysiologic, and behavioral data suggest that IMPs, composed of an FDA-approved material, are a strong candidate for acute pharmacologic intervention in human TBI. IMP treatment in experimental TBI not only preserved brain matter and reduced edema but also preserved function. Validation of the effects of IMP in 2 different TBI models supports the potential benefits in the heterogeneous injury seen in human TBI. IMPs are portable and stable at room temperature and thus could be dispensed easily by first-responder teams or in the emergency room. Our findings suggest that IMP potentially could become a clinically effective pharmacologic therapy for TBI.