Assessment of the Effects of Stretch-Injury on Primary Rat Microglia

Mechanical stretch-injury is a prominent force involved in the etiology of traumatic brain injury (TBI). It is known to directly cause damage and dysfunction in neurons, astrocytes, and endothelial cells. However, the deleterious effects of stretch-injury on microglia, the brain’s primary immunocompetent cell, are currently unknown. The Cell Injury Controller II (CICII), a validated cellular neurotrauma model, was used to induce a mechanical stretch-injury in primary rat microglia. Statistical analysis utilized Student’s t test and one- and two-way ANOVAs with Tukey’s and Sidak’s multiple comparisons, respectively. Cells exposed to stretch-injury showed no signs of membrane permeability, necrosis, or apoptosis, as measured by media-derived lactate dehydrogenase (LDH) and cleaved-caspase 3 immunocytochemistry, respectively. Interestingly, injured cells displayed a functional deficit in nitric oxide production (NO), identified by media assay and immunocytochemistry, at 6, 12, 18, and 48 h post-injury. Furthermore, gene expression analysis revealed the expression of inflammatory cytokines IL-6 and IL-10, and enzyme arginase-1 was significantly downregulated at 12 h post-injury. Time course evaluation of migration, using a cell exclusion zone assay, showed stretch-injured cells display decreased migration into the exclusion zone at 48- and 72-h post-stretch. Lastly, coinciding with the functional immune deficits was a significant change in morphology, with process length decreasing and cell diameter increasing following an injury at 12 h. Taken together, the data demonstrate that stretch-injury produces significant alterations in microglial function, which may have a marked impact on their response to injury or their interaction with other cells.


Background
Traumatic brain injury (TBI) is a growing severe public and military health concern, with 1.7 million cases accruing an economic burden near 76.8 billion annually [1]. TBI is divided into two injury phases, primary and secondary [2]. The primary injury results from the impact or acceleration events generating mechanical forces that act upon the sensitive brain tissue, resulting in immediate, irreparable damage, and dysfunction. Secondary injury processes resulting from the initial injury evolve and contribute to the destruction of brain tissue. A prominent secondary injury process is the inflammatory response, primarily mediated by activated microglia, the brain's immunocompetent cell [3]. The initial immune response is thought to be a protective and restorative process; however, there are multiple experimental models [4][5][6][7] and human [8,9] TBI cases that implicate prolonged microglia activation in harmful outcomes. Currently, the mechanism causing prolonged microglial activation is unknown.
Mechanical strain produced from the primary injury generates shearing forces that stretch and deform the cytoarchitecture and cells, contributing to death and dysfunction. Stretch-injury in vitro models allows for the study of the effects of primary injury forces on individual cells. To date, in vitro models of neurotrauma show neuronal [10,11], astrocyte [12], and endothelial [13] pathologies that mimic those injuries observed in in vivo TBI models. For example, rat primary cortical neurons exposed to a uniaxial stretch-injury with a strain of 10% displayed focal axonal swellings, similar to those observed in diffuse axonal injury (DAI) in experimental models and human cases of TBI [10]. Human astrocytes exposed to a biaxial stretch-injury with a strain of 20% displayed retracted end-feet processes and activation c-Jun NH 2 -terminal kinase (JNK), replicating astrocyte associated dysfunction in blood-brain barrier (BBB) integrity. Furthermore, murine endothelial cells exposed to a stretchinjury with a 20% strain displayed markedly retracted, swollen and deformed morphology with notable intercellular spaces between cells [13]. To date, the effects of stretch-injury on microglia viability, phenotype, and function are currently unknown.
Following the initial trauma, microglia interact with trauma associated-factors and become activated; this transformative process is necessary to mediate an inflammatory response. This is characterized by a morphological and phenotypic transformation known as activation [14]. The morphological changes involve a transition from a ramified to amoeboid shape, and functional changes such as upregulation of nitric oxide (NO) [3] and reactive oxygen species (ROS) production [14], the release of chemokine/cytokines [2], alteration in cell surface immune receptors and functional enzymes, and an increase in phagocytic and migratory behavior [15]. Persistent, or chronic activation of microglia following a TBI can occur if the initial response is left unresolved, leading to functional deficits and neurodegeneration [16]. To date, the mechanism(s) underlying chronic post-traumatic microglia activation remain elusive, and given that the primary insult can induce cellular pathologies in other cell types, an investigation into the effects of primary forces on microglia function is warranted.
In this study, we aimed to evaluate the impact of stretchinjury on microglial viability, function, and morphology. Additionally, the effect of stretch-injury on microglial priming was examined by examining microglia viability and function after microglia were treated with a known inflammogen after a stretch-injury. First, time-course studies were conducted using media assays to assess viability (LDH) and function (NO). From this, we focused on a single post-injury time point to evaluate inflammatory-associated cytokines and enzymes, reactive oxygen species, and morphology analysis. Lastly, a time-course study assessed the priming effect of stretchinjury on microglial viability and function. This study is the first to comprehensively study the effect of stretch-injury on microglial viability, morphology, and function.

Stretch-Injury
Cells were seeded at a density of 3 × 10 5 cells/mL on Pronectin-coated silicone-based deformable membrane BIOFLEX® culture plates (BF-3001P Flexcell International, Hillsborough, NC). Cells were incubated for 24 h in growth media to allow for adherence (37°C, 5% CO 2 ), after which a stretch-injury was administered, as previously described [11]. Briefly, a stretch-injury was produced using the Cell Injury Controller II (CIC II), which utilizes nitrogen gas to pressurize a BIOFLEX® well individually. The CIC II regulates the duration and pressure of gas delivered to the well by using a rubber tube that feeds through a plastic circular well adaptor plug that attaches to the top of the well, creating a hermetically sealed system. Upon discharging the gas, the well becomes pressurized, and a bi-axial deformation of the flexible membrane, resulting in the cultured cells experiencing a stretchinjury. All stretch-injury experiments used identical parameters with 60 psi regulator pressure for a duration of 50 mSec, resulting in a 5 psi pressurization of the well that created a 20% stretch event in the center of the well. The parameters used for this study were determined by a stretch-injury doseresponse pilot study that established the CICII's highest setting, 60 psi, resulted in functional deficits (data not shown).

Lipopolysaccharide (LPS) and Etoposide Treatment
The potential priming effect of stretch-injury was evaluated by treating primary microglia with 100 ng/mL concentration of LPS (Sigma, L8274) 15 min after stretch-injury [3]. Additionally, an LPS treated-only group served as an activation control group. A dose of 15 uM of Etoposide (catalog no. E1383; Sigma) was used to induce apoptosis and elicit a robust cleaved caspase-3 response to serve as a positive control [18].
Absorbance at 496 nm was measured using a ChroMate microplate reader (Awareness Technology Inc., Palm City, FL).

Nitric Oxide (NO) Measurement
In conjunction with the cell death assay, media NO was measured in primary microglia exposed to stretch-injury at 1, 6,12,18,24,48, and 72 h post-injury. NO content in the media was quantified using a Griess reagent assay kit (Invitrogen). The assay was performed per the manufacturer's recommendations and as previously described [19]. Colorimetric changes in 96-well plates were quantified with a Chroma Plate Reader (Midwest Scientific, St. Louis, MO, USA) at 545 nm.

Immunocytochemistry
Cells were fixed with 4% paraformaldehyde and rinsed twice with 1× PBS. Primary antibodies included rabbit antiarginase-1 (20 μl/ml; Abcam), rabbit anti-NOS (10 μl/ml; Abcam, Cambridge, MA) and rabbit anti-cleaved caspase-3 (10 μl/ml; Abcam). Also, cell nuclei were stained with DAPI (0.5 μl/ml; Life Technologies, Carlsbad, CA). Fluorescently tagged secondary antibodies (Invitrogen) were visualized with an Olympus BX43 fluorescent microscope with a CellSens Standard imaging program. Five randomly selected images were taken per sample, and each image was qualitatively assessed for pixel density and normalized to the total number of DAPI stained cells. Images were taken from the center of the membrane.

Protein Quantification
Protein was collected from cells at 12 h post-injury. Cells were lysed, treated with protein inhibitor (Halt Protease Inhibitor single use cocktail, Thermo Scientific) then centrifuged.
The supernatant was collected, and western blot analysis was performed as previously described [20]. Briefly, 25 μg of protein sample was run on a Mini-protean TGX precast gel then transferred to a nitrocellulose sheet (Trans-Blot Turbo Transfer System, Biorad, Hercules, CA, USA). The nitrocellulose sheet was blocked for 1 h then incubated overnight with the following antibodies: 3-NT (1:1000, Abcam) and gp91(1:5000, Abcam). Resultant bands were quantified with Image J densitometry and normalized to a loading control protein, beta-actin (1:5000, Abcam).

RNA Isolation and Gene Analysis
TRIzol® (Invitrogen, Carlsbad, CA) was used to extract mRNA from the cells, and mRNA concentrations were measured by a Nanodrop system (Thermo Scientific). A Veriti thermal cycler (Applied Biosciences) and a complementary DNA (cDNA) conversion kit (Applied Biosciences, Waltham, MA) was used to convert 1 μg of mRNA into cDNA according to the manufacturer's suggested protocol. All quantitative real-time PCR (qRT-PCR) data was procured through the StepOnePlus Real-Time PCR System (Applied Biosciences), according to the manufacturer's recommended protocol. All primers were designed by Sigma-Aldrich (Darmstadt, Germany) and listed in Table 1.

Migration Assay and Morphological Evaluation
A cellular exclusion zone assay was used to quantify migration. Cells were seeded at a density of 3 × 10 5 /mL on Pronectin-coated BioFlex well plates with a 1.4-cm × .1-cm strip of silicone membrane placed in the center of the well. A graded grid was drawn onto the underside of the membrane as a reference. Microphotographs were taken using a phasecontrast microscope (Zeiss, PrimoVert) with a USB adapted ocular (Swiftcam 5.0-megapixel digital camera microscopes, Amazon.com) for capturing images. Images were captured at pre-injury baseline and 12, 24, 48, and 72 h post-injury. ImageJ (NIH, open-access software) was used to quantify cell exclusion zone diameter. Twenty-four hours after plating, the barrier strip of silicone was removed, and a baseline image was taken to define the cell exclusion zone. Then a stretchinjury was administered, and cells were returned to the incubator. At pre-stretch, 12, 24, 48, and 72 h post-injury time points, cells were counted in the exclusion zone. Cells were counted if they matched these criteria: (1) dark coloring and (2) circular or branched phenotype. In some experiments, cells were able to attach underneath the silicone barrier in the exclusion zone. These cells were counted and subtracted from the total number of migrating cells in subsequent post-injury assessment time points, 12, 24, 48, and 72 h.
Morphology was evaluated using the same devices, and images were taken 12 h post-injury. ImageJ (NIH, openaccess software) was used to quantify cellular characteristics, i.e., number and length of processes and cell diameter.

Phagocytosis Assay
At 12 h after stretch-injury, cells were assessed for phagocytic activity using the Phagocytosis Assay Kit (Cayman Chemical, Ann Arbor, MI). Briefly, at 8 h after injury, latex beads with rabbit IgG-FITC conjugates (1:500) were incubated with control and injured cells for 4 h, followed by a 1-min incubation with trypan blue to quench non-phagocytosed bead fluorescence and cell fixation with 4% paraformaldehyde. Cells were counterstained with DAPI, and five images at × 20 magnification were collected from each coverslip. The number of DAPI cells containing FITC-labeled beads were counted and normalized to the total number of DAPI. This assay was repeated in triplicate.

Statistical Analysis
Statistical analysis was conducted using Graphpad Prism Software version 9.01. (GraphPad Software, San Diego, CA). NO and LDH assays were performed in triplicate for each trial, and the trials were replicated four times to generate a sample size of n = 4 per group. A two-way ANOVA with Sidak's multiple comparisons test was conducted on the groups in which controls were set to 1.0. The stretch-injured values were normalized to control values for each trial for NO and LDH assay measures. ROS assays were conducted singly for each trial, and the trials were replicated three times to generate a sample size of n = 3. A one-way ANOVA with Tukey's multiple comparisons test was conducted. Western blot analysis was performed in pooled samples from triplicates for each trial, and the trials were replicated three times to generate a sample size of n = 3. Immunocytochemistry experiments were conducted singly and repeated three times to create a sample size of n = 3. Western blot and immunocytochemistry results were analyzed using an unpaired Student's t test. The migration assay was conducted singly for each trial and replicated three times for an n = 3. The total number of migrating injured cells was normalized to control cells at each time point. A two-way ANOVA with Sidak's multiple comparisons test was conducted for assay measures. The morphology analysis was assessed from five images per trial, and each trial was replicated three times to generate a sample size of n = 3. A total of 15 images were analyzed for the control and stretch-injured condition. Percent averages for each group were assessed using a two-way ANOVA with Sidak's multiple comparisons. For all statistical tests described, a p value < 0.05 was considered statistically significant. Data are presented as mean ± standard deviation (SD).

Stretch-Injury Does Not Increase LDH at Multiple Time Points or Cleaved Caspase-3 at 12 h Post-Injury
A time-course study of media-derived LDH was conducted to determine the effect of stretch-injury on microglial viability and membrane permeability at acute and long-term time points. LDH is typically confined in the cytosol and is detectable in the event of cell lysis, necrosis, or compromised membrane permeability. Following a 20% stretch-injury, media was assayed for LDH at 1, 6, 12, 18, 24, 48, and 72 h postinjury and showed no significant differences in media LDH content between stretch-injured and control cells at any time point (Fig. 1a). To further assess microglia viability, an immunocytochemistry evaluation of cleaved caspase-3, a marker of apoptosis, was conducted on fixed cells at the 12-h postinjury time-point and revealed no significant difference between injured and non-injured cells (Fig 1b), the observation that LDH media levels were unchanged between controls, and injured cells suggests that a 20% stretch-injury does not alter microglia viability or membrane permeability. Furthermore, there was no significant difference between control and injured cells of cleaved caspase-3 expression, a marker of apoptosis, demonstrating that 20% stretch-injury does not induce apoptosis. Taken together, these results suggest that a 20% stretch-injury does not significantly affect microglia viability or membrane permeability at 12-h post-injury.

NO Release and iNOS Expression Is Significantly Downregulated Following Stretch-Injury
To examine the acute and long-term effect of stretch-injury on microglia function, a time course study, using assessment of NO in the media, was conducted at 1, 6, 12, 18, 24, 48, and 72 h post-injury. Stretch-injury significantly downregulated release of NO at 6, 12, 18, and 48 h post-injury ( Fig. 2a; 6, 12, and 18 h, p < 0.05; and 48 h, p < 0.01, two-way ANOVA, Sidak's multiple comparisons post-test). Immunocytochemistry evaluated the primary enzyme responsible for catalyzing NO production, iNOS, at 12 h post-injury and showed that stretch-injury significantly decreased iNOS expression compared to nonstretched control cells ( Fig. 2b; p < 0.01 unpaired Student's t test). Compared to controls, stretch-injury did not alter gene expression of iNOS at 12 h post-injury. The observed decreases in NO release and iNOS protein expression following a 20% stretch-injury suggest that a mechanical stimulus can reduce microglia function and that this reduction is not related to a decrease in cellular viability

Oxidative Stress is Unaffected by Stretch-Injury
ROS production was assessed at 12 and 24 h post-stretch to determine if this microglial activation product and subsequent oxidative stress aligned with NO release. Evaluation of ROS at 12 and 24 h post-injury showed that there is no significant effect of stretch-injury on ROS production at either time point compared to control cells (Fig. 3a). There was no effect of stretch-injury on microglia-mediated oxidative stress, Fig. 1 Stretch-injury does not affect primary microglia viability. a Primary microglia exposed to a stretch-injury of 5 psi show no significant difference in media LDH when compared to control cells. Absorption values of the injured condition were normalized to controls (set as 1.0). The same cells were sampled over the entire time-course study. b Representative sample of one trial presented. Scale bar is 50 μm. c Immunocytochemistry evaluation of cleaved caspase-3 indicates no significant difference between injured and control cells. Points represent individual data points. Pixel density was normalized to DAPI cell count. N = 3/group. Error bar represents ± SD evaluated using western blot analysis for 3-NT, at 12 h postinjury (Fig. 3b). To further investigate ROS production, the expression of the NOX2 membrane-bound enzyme component, gp91, was evaluated 12 h post-injury using western blot. Its expression, compared to control cells, was not significantly altered (Fig. 3c). The data shows that ROS production and oxidative stress do not track with decreases in NO release.
Genes Associated with Microglia Activation Are Significantly Downregulated or Unaffected Following Exposure to Stretch-Injury The expression of genes associated with microglial activation in TBI was evaluated using RT-PCR at 12 h postinjury. The 12-h time point was chosen based on preliminary data identifying it as the first time point after postinjury to display a significant decrease in microglia function, NO release. Pro-inflammatory associated genes, IL-6 [4], CD86 [21], and IL-1 β [4], were selected for their observed upregulation in activated microglia in the secondary injury cascade in experimental models of TBI. Antiinflammatory-associated genes were chosen for their microglia-associated roles in TBI: immunomodulation (IL-10) [22], tissue repair (arginase-1) [23], homeostatic maintenance (TGFβ) [24], and pathogen recognition and capture (CD206) [2]. Pro-inflammatory-associated cytokine gene IL-6 was significantly downregulated compared to controls, 12 h after stretch-injury ( Fig. 4e; p < 0.05 unpaired Student's t test). Interestingly, the antiinflammatory-associated gene coding for the enzyme arginase-1 was also significantly downregulated at the 12-h post-injury time point (Fig. 4a p < 0.05 unpaired Student's t test). Immunocytochemistry analysis of arginase-1 confirmed the genetic findings ( Fig. 4i; p < 0.05 unpaired Student's t test). There was no significant difference in gene expression of pro-inflammatoryassociated genes, CD86, and IL-1β (Fig. 4f, g) or antiinflammatory-associated genes CD206, IL-10, and TGFβ

Morphology Is Significantly Altered in Response to Stretch-Injury
Following a TBI, microglia undergo a morphological transformation from a small, ramified cell body to a broader, rounded amoeboid phenotype. The effect of mechanical strain on morphology was evaluated using phase-contrast microscopy; images were taken from the center of the well 12 h postinjury. The cell count, cell body area, number, and length of processes were evaluated using Image J analysis software. Following stretch-injury, microglia display a significant increase in their cell body area, which is defined by the total area of the cell except for their processes, compared to controls at 12 h after injury (Fig. 5d, p < .01, unpaired Student's t test). Furthermore, stretch-injured cell processes length is significantly decreased compared to control cells 12 h following injury. (Fig. 5e, p < .05, unpaired Student's t test)

Migration Is Impaired Substantially Following Stretch-Injury
In response to injury, microglia localizes to the site to help contain the damage and mediate the immune response. To assess this behavior following injury, we developed and conducted a longitudinal examination of migratory behavior using a cell exclusion zone assay. The results showed significantly fewer cells migrated into the exclusion zone in the injured condition at 48 and 72 h after injury than control cells (Fig. 6, p < .001 and p < .0001, two-way ANOVA with Sidak's multiple comparisons). The exclusion zone areas from each plate were quantified. They showed no significant difference between groups (p = 0.687, unpaired t test), with an average area of 0.551785 mm 2 ± .044266 for the controls and an average area of 0.568224 mm 2 ± .045897 for the injured group.

Phagocytosis Is Not Affected by Injury
Phagocytosis is a critical microglial function that aids in removing dead debris and damaged cells following injury. Interestingly, there was no difference between bead uptake between the control and injured cells, demonstrating that stretch does not affect phagocytosis 12 h after injury. An LPS treated (100 ng/mL) was included as a positive control. Briefly, cells were incubated with LPS for 44 h and treated with identical conditions as the control and injured cells. The LPS-treated group displayed a robust trend towards increased bead uptake (Fig. 7, p = .0501, one-way ANOVA with multiple comparisons) compared to controls.

Stretch-Injury Does Not Prime nor Depress Microglial NO Release nor Alter LDH Release When Exposed to a Subsequent LPS Challenge
To evaluate the potential priming effect of stretch-injury, primary microglia were exposed to a stretch-injury then treated with LPS (100 ng/mL) 15 min after the injury. Following treatment, cells were assayed for viability and function using LDH and NO media assays, respectively. Compared to LPS treated-only cells, stretch-injury did not significantly alter viability when subsequently treated with LPS at 24, 48, or 72 h post-injury (Fig. 8a). There was a significant increase in media LDH in both the LPS treated only and LPS plus stretch-injury at 24, 48, and 72 (Fig. 8a, p < .001, p < .001, and p < .05) hours compared to control cells. Stretch-injury does not prime or depress NO release when treated with LPS 15 min after injury compared to LPS treated only cells at 24, 48, or 72 h post-injury (Fig. 8b). Cells that were exposed to a stretchinjury plus LPS and LPS treated only displayed a significant increase in NO release at 48 (Fig. 8b,   to determine if microglial responses in the pro-inflammatory condition mimic those observed in TBI, gene expression of select inflammatory associated genes was evaluated in the pro-inflammatory group at 20 h post-injury. Cells treated with LPS, both with and without a stretch-injury, displayed a significant increase in iNOS and IL-1β (Fig. 8, p < .05 and p < .01, respectively. One-way ANOVA with Dunnett's post-test) and a significant decrease in arginase-1, IL-10, and TGF-β (Fig. 8d, p < .0001, p < .05, and p < .05, respectively. Oneway ANOVA with Dunnett's post-test). Interestingly, the trend of inflammatory gene expression following injury was conserved at 24 h with IL-6, arginase-1, IL-10, and TGF-β expression significantly decreased (Fig. 8c, d, p < .0001, p < 0001, p < .01, and p < .05, respectively, one-way ANOVA with Dunnett's post-test).

Discussion
The findings presented here address a significant gap in the microglia literature in the context of TBI, specifically the consequences of the primary-injury on microglia viability, morphology, and function. We demonstrate that rat primary microglia remain viable while displaying significant functional and morphological perturbations in response to a stretchinjury with a 20% strain magnitude. The longitudinal assessment of media LDH from stretchinjured cells showed no significant changes at any post-injury time point examined compared to controls, demonstrating that stretch-injury does not result in altered membrane integrity or necrosis. Media levels of the intracellular enzyme LDH indicate compromised membrane integrity and necrosis [25,26]. For example, stretch-injured rat cortical astrocytes exposed to injury parameters similar to those used in this study exhibit increased media LDH up to 24 h after injury, suggestive of injury-induced membrane permeability [26]. Additionally, stretch-injury does not induce apoptosis, as indicated by no changes in cleaved caspase-3 expression between injured and control cells. These findings are counter to what is found in stretch-injured astrocytes [12] and neurons [27]. They can be attributed to differences in cellular size, morphology, and functional roles in the brain. Astrocytes are significantly larger than microglia and, as a consequence, have more surface area exposed to the strain forces, leading to increased membrane alterations [28]. Similar to astrocytes, neurons vary significantly in cell diameter and morphology compared to microglia, with a larger cell body and long, delicate axons that are susceptible to damaging strain forces [10]. Furthermore, microglia may be less sensitive to harmful strain forces than astrocytes and neurons because they are constantly surveying the brain through migratory and motile behavior. In contrast, astrocytes and neurons are adherents, stationary, and nonmigratory in non-pathological conditions, rendering them more susceptible to mechanical deformations. Taken together, these results suggest that stretch-injury does not alter viability or membrane permeability.
Stretch-injury significantly downregulated iNOS expression and decreased media levels of NO. These findings d Injured cells display a significantly larger cell body area compared to controls. e The average process length is significantly decreased compared to controls. N = 3/group. All groups were compared using unpaired t test, * p < .05, **p < .01. Bars represent mean ± SD Fig. 7 Stretch-injury does not alter phagocytic activity. a Representative images of latex beads conjugated with FITC-IgG (green) were incubated for 4 h with control, stretch-injured, and LPS-treated primary microglial cells. Scale bar set to 50 μm. b Quantitation of phagocytosed beads demonstrated that stretch-injury did not alter phagocytic activity compared to control cells at 12 h post-injury (8 h following stretch-injury cells were treated with FITC-IgG for 4 h). The LPS positive control (100 ng/mL) displayed a strong trend towards phagocytosing significantly more beads when assessed at 48 h post-treatment (44 h after LPS treatment cells were incubated with FITC-IgG for 4 h). For quantification, FITC-IgG beads were normalized to total DAPI cell count contrast to what is observed in patients and experimental models of TBI. They can be attributed to fundamental modeling differences and may indicate an innate response of microglia to the injury itself. To assess the effects of primary injurious forces on microglia, mono cultured are exposed to damage in a sterile environment, free from other damaged cells and damage-associated molecular patterns (DAMPS). Because of this unique environment, microglia activation and subsequent upregulation of iNOS and NO production are not observed. Microglia upregulate the immune-response enzyme iNOS increasing the production of NO, an inflammatory mediator. Multiple experimental models of TBI [29][30][31][32] show a significant upregulation of iNOS following injury. Continuous production of NO, through increased iNOS expression and activity, can lead to cytotoxic oxidants, such as peroxynitrite, that contribute to oxidative stress [20,33,34]. Multiple aspects of the process involved in NO production were evaluated in the current study. Media NO levels were significantly downregulated at 6, 12, 18, and 48 h after injury compared to uninjured cells.
Consistent with the decreased NO in the media, iNOS protein expression was significantly downregulated 12 h following injury. However, mRNA transcripts were unaffected by injury at the same time points compared to non-stretched cells.
Interestingly, a rat model using a controlled cortical impact (CCI) showed that tissue NO is significantly decreased at 2, 6, and 12 h following injury near the injury site. However, opposite to our iNOS findings, there was a significant increase in NADPH-diaphorase-positive cells, indicative of NOS activity, in the same regions where the NO decrease was detected. The authors attributed the reduction in NO to it being converted into peroxynitrite [35]. Additionally, consistent with the findings of the in vivo study, we observed a significant decrease in media NO levels following injury; however, in contrast to the animal model, we found that the protein expression of iNOS is significantly decreased. This can be explained by the context of the in vitro environment, which lacks damaged associated factors that lead to microglial activation and increase iNOS expression. Future studies will evaluate the mechanism underlying the marked reduction in iNOS expression and determine if stretch-injury alters NO production in activated microglia. Our findings show that a stretch-injury suppresses media NO levels and downregulates iNOS protein expression, suggesting that basal level NO production is affected by stretch-injury.
The finding that stretch alters media NO and iNOS expression, but not iNOS mRNA transcription, is puzzling. A possible explanation could be that stretch-injury disrupts protein synthesis by inducing endoplasmic reticulum stress, observed in blast [36] and fluid percussion injury [37] animal models of TBI. Additionally, rats exposed to a CCI model of TBI showed differentially methylated mRNA transcripts [38], a prominent post-transcriptional modifier of RNA, and has been shown to inhibit translation [39]. Collectively, the data indicate that stretch-injury significantly downregulated NO media production at multiple time points, and this is mediated by a significant decrease in iNOS expression, which demonstrates that stretch-injury produces functional deficits in immune function.
Experimental models of TBI show that chronically activated microglia contribute to neuronal damage and dysfunction by producing free radical oxygen species that lead to cytotoxic oxidative stress [5,20,40]. Following activation, microglia upregulate NOX2 and increase the production of ROS, that when combined with RNS, can form cytotoxic factors, such as peroxynitrite, that lead to protein nitrosylation resulting in cellular dysfunction or death. To evaluate the effect of stretch-injury on microglial ROS generation and oxidative stress, ROS assessments, oxidative stress, and the NOX2 enzyme were performed. Production of ROS was conducted by evaluating media and cellular levels 12 h after injury. Interestingly, there was no effect of stretch-injury on ROS production compared to control cells. Next, western blot analysis showed that the expression of gp91, the membrane-bound component of the NOX2 complex, was unaffected by stretch-injury. Lastly, oxidative stress was evaluated by western blot for 3-NT, a marker of protein nitrosylation, specifically tyrosine residues. Consistent with the other ROSrelated outcome measures, stretch-injury did not affect the oxidative stress response, specifically, protein nitrosylation, indicated by no changes between control and injured cells using western blot analysis of 3-NT. These results are surprising, given that multiple lines of research show both NO and ROS production are increased following microglial activation [41][42][43]. Contrary to the RNS data, ROS does not appear to be affected by stretchinjury.
Following a TBI, activated microglia release both pro and anti-inflammatory-associated cytokines [44]. Multiple converging lines of research show that the post-insult inflammatory response in the brain presents with a pro and antiinflammatory phenotype, and this duality varies temporally. For example, a rodent model of diffuse brain injury, where strain-induced injuries are prevalent, showed an increase in both pro (IL-1β, CD14 and iNOS) and anti (arginase-1) inflammatory mRNA transcripts at 24 h post-injury [29]. The effect(s) of the primary forces on the expression of microglial inflammatory-related genes is undefined. To address this gap in the literature, mRNA transcripts of inflammatoryassociated genes were evaluated using RT-PCR at 12 h poststretch-injury. Interestingly, we found that both proinflammatory (IL-6) and anti-inflammatory (IL-10 and arginase1) associated genes were downregulated following injury. This is counter to what is observed in experimental models and TBI patients, where these genes are commonly upregulated. For example, plasma levels of IL-6 have significantly elevated acutely in patients who sustain an mTBI [45]. In a rat model of experimental TBI, there was an immediate increase in brain IL-10 mRNA transcripts, peaking at 4 h postinjury and continuing to rise until 20 h post-injury [46]. In a mouse model of TBI, arginase-1 mRNA transcripts were significantly upregulated 24 h following injury in both young (3month old) and old (24-month old) animals [30]. However, an important distinction between our findings and others is (1) monoculture microglia were exposed to the stretch-injury, and (2) they do not interact with DAMPs or damaged cells and do not appear to be activated, either morphologically or functionally. Therefore, the upregulation of inflammatory genes would not be expected. Stretch-injury suppresses multiple genes associated with the inflammatory response following injury; the consequence and long-term ramification of this effect on microglial inflammatory response is an important question since dysfunctional inflammatory responses are observed in the chronically injured brain. Stretch-injury decreases basal level microglial migration. Following the initial injury, microglia sense factors from the trauma and migrate to the injury site to remove debris and damaged cells [47]. To date, the effect of stretch-injury on migration in the biomechanically injured brain is unexplored in the literature. To address this gap in the literature, we assessed migration following stretch-injury by using a cell exclusion assay and found that stretch-injury significantly reduced the number of cells that migrated into the exclusion zone at 48 and 72 h post-injury. These findings show that a mechanical stimulus can reduce an essential functional behavior integral to the response to injury and necessary for homeostatic function. Future studies will elucidate the extent these migratory deficits play in a physiologically relevant setting.
Microglial morphology is significantly altered following exposure to stretch-injury. Following the primary injury, microglia undergo a morphological transformation characterized by a significant increase in soma diameter and reduction of processes to take on an amoeboid phenotype [48]. Phase-contrast imaging showed that stretch-injury significantly reduced process length 12 h after injury, demonstrating that a deformation event alone can rearrange the cytoskeleton. This is consistent with morphological alteration in neurons [49,50], astrocytes [51], and endothelial [13] cells exposed to stretch-injury. The stretch-induced reduction in process length may contribute to the decreased number of cells migrated into the exclusion zone. Since migration involves remodeling of the cytoskeleton, one explanation could be that stretch-injury alters the cytoskeleton and impairs its remodeling during migration.
The microglial inflammatory response is not altered following treatment with LPS post-stretch-injury. To extend our model into a more physiologically relevant setting, where mechanical forces are followed by exposure to inflammatory challenges, stretch-injured microglia were treated with LPS, a potent microglial activator, 15 min following injury. Then media levels of NO and LDH were assessed at 24, 48, and 72 h post-treatment to determine viability and activation. Surprisingly, the suppressive effect of stretch-injury on NO media levels was not observed in the post-stretch, LPS treated cells. Consistent with the LDH results from the time course assessment, LDH media levels were unchanged between any groups. The unexpected NO media result could be attributable to multiple issues. The first could be the potency of LPS as an inflammatory challenge.
Given that it is an endotoxin found on the outer membrane of gram-negative bacteria, it is expected that elicits a more robust inflammatory response than factors that are released during a TBI. For example, primary rat microglia treated with interferon-γ (IFNγ) and tumor necrosis factor-α (TNFα), factors associated with brain trauma, display a varied inflammatory response compared to LPS. Compared to the IFNγ and TNFα treatment, LPS-treated cells expressed significantly more NOS and arginase-1 mRNA transcripts and protein expression. Additionally, media NO levels were also considerably higher. The second issue could be the timing of the challenge post-injury. The effects of stretch-injury may not have enough time to alter the response to LPS. Future studies will challenge stretchinjured cells at immediate and delayed time points. These results suggest that stretch-injury does not change microglial inflammatory response when subsequently challenged; however, as noted above, multiple factors could account for these findings.
Phagocytosis is unaffected following stretch-injury. Multiple studies show that activated microglia significantly increase their phagocytic function compared to their resting cells [52][53][54]. A critical role of activated microglia in the postinjured brain is phagocytosing debris and dead and damaged cells [23]. The phagocytosis assessment on stretch-injured cells showed no difference in fluorescently label conjugated beads uptake between the control and stretch-injured group, 12 h post-injury. Given that basal level phagocytosis in resting microglia is already low, it was not expected to see an effect of a stretch injury. An interesting observation was the relatively low number of phagocytosed beads in the non-activated cells cultured on Pronectin. This finding is consistent with a study showing that when primary rat microglia are incubated with fibronectin, there is a reduction in phagocytosis, as measured by the engulfment of zymogen particles. Furthermore, the same study showed that morphological phenotypes were consistent with our control cells, with cells displaying a reduced cell body diameter and an increased number of processes [55].

Conclusions
The results presented here are the first to examine the consequences of the primary injury on microglia by characterizing the effects of stretch-injury on viability, morphology, and Fig. 8 Stretch-injury does not alter microglia NO or LDH responses to subsequent LPS stimulation. Stretch-injury before LPS treatment does not induce a significant difference in LDH release between LPS-treated and stretch-injury plus LPS treatment. Both LPS and LPS plus stretchinjury significantly increase LDH release at a 24, b 48, and c 72 h posttreatment. Pre-stretch-injury does not alter NO responses when subsequently exposed to LPS. LPS and stretch-injury plus LPS both significantly upregulate NO at e 48 and f 72 h post-treatment. Treatment with LPS and LPS plus injury significantly increases g iNOS and i IL-1β and decreases h IL-6, j arginase-1, k IL-10, and l TGF-β gene expression at 24 h after injury. There was no difference between the LPS and LPS plus injury groups. N = 3/group. One-way ANOVA with Dunnett's post-test. *p < 0.05, **p < 0.01, ***p < 0.001,****p < 0.0001. Experimental groups are normalized to the control group (a, b). For gene expression analysis, all experimental group changes are represented as fold changes compared to the control group. Error bars represent mean ± SD function. The data shows that microglia remain viable after the injury and display morphological alterations, with decreased process length and functional deficits, with reduced migration and nitric oxide production, at 12 h after injury. Additionally, stretch-injury significantly downregulates inflammatoryassociated gene transcripts at 12 h. These findings will contribute to the understanding of microglia in the context of TBI, and future studies will elucidate the molecular mechanism(s) that are responsible for these deficits. Furthermore, studies will examine the contributions of these effects of dysfunctional microglia response observed in experimental models and individuals with TBI.