TNF-α Impairs Pericyte-Mediated Cerebral Microcirculation via the NF-κB/iNOS Axis After Experimental Traumatic Brain Injury

Secondary structural and functional abnormalities of the neurovascular unit are important pathological mechanisms following traumatic brain injury (TBI). The tumor necrosis factor α (TNF-α)/nuclear factor-κB (NF-κB) pathway regulates neuroinammation and oxidative damage, which may act as triggers for pathological processes after TBI. However, the role of TNF-α/NF-κB in pericyte-mediated cerebral microcirculation are currently unknown. 0.01.

this led to pericyte-mediated disturbance of the cerebral microcirculation, which may be one of the vital mechanisms of secondary injury in TBI.

Background
Secondary injury induced by traumatic brain injury (TBI) is considered to be a reversible pathological process, which involves vascular, cellular, metabolic, molecular, and other factors, of which neuroin ammation and oxidative stress are the key links [1][2][3]. In the past, TBI research focused on neurons, ignoring the role of glia, and the cerebrovascular system [4]. After the concept of the neurovascular unit (NVU) was proposed, people gradually realized that neurons, blood vessels, and glia as a whole maintain homeostasis and the function of the central nervous system (CNS) through crosstalk [5][6][7]. The conceptual framework of the NVU may contribute to understanding the pathophysiological changes of TBI. Recent studies have demonstrated that the NVU is a key participant in secondary injury following TBI [8,9].
The NVU is composed of neurons, microglia, astrocytes, pericytes, smooth muscle cells, and endothelial cells. It is the minimum acting unit necessary to maintain the proper function of the CNS [10]. The NVU maintains blood brain barrier (BBB) and vascular integrity through the interaction between its constituent cells [11]. After TBI, microvascular dysfunction is mainly manifested by NVU dysfunction, which is regulated by pericyte function [12]. Pericytes are microvascular mural cells distributed in precapillary arteries, capillaries, and postcapillary venules [13]. In the NVU, pericytes interact with neighboring cells and process signals to execute multiple functional responses, particularly to the regulation of the microvessels [14]. Recent studies have shown that pericyte degeneration after TBI led to regional microcirculatory hypoperfusion and an increase of BBB permeability, which mediated microcirculation disturbance [15][16][17]. A deeper exploration of pericyte-mediated microcirculation disorders is essential to investigate the pathological mechanisms following TBI.
An increasing body of evidence shows that the pericyte-mediated microcirculation function is in uenced by neuroin ammation [18,19]. Microglia are the major immunocompetent cells in the NVU, and their activation is a hallmark of neuroin ammation [6,20]. After the initial injury, the altered microenvironment and intracellular components released from damaged cells trigger local glial activation and recruitment [5]. Activated microglia induce neuroin ammation and aggravate tissue damage by promoting the release of in ammatory factors, in ammatory cell in ltration, as well as free radical production [21,22]. The generated free radicals not only induce oxidative stress but also act as an in ammatory mediator to amplify in ammation [2]. In the brain, a large pool of microglia (also known as perivascular microglia) are located at the proximal region surrounding the cerebrovasculature. Moreover, confocal laser scanning microscopy analysis revealed the highest density of microglial endfeet contacting the glial basement membrane around capillaries, with signi cantly lower densities around arteries and veins. These structural features allow a close interaction between microglia and pericytes. Indeed, many studies have demonstrated a very tight spatiotemporal correlation between vascular activation, cerebral blood ow (CBF) restriction, BBB breakdown, and activation of brain-resident microglia [23][24][25].
Under pathological conditions, the in ammation-related tumor necrosis factor α (TNF-α)/nuclear factor-κB (NF-κB) pathway is activated, which is closely related to the prognosis of TBI [26,27]. TNF-α activates IKK-α/β by binding to its receptor, causing IκB phosphorylation and promoting NF-κB nuclear translocation, which in turn regulates NF-κB-related signaling pathways [27]. As a pro-in ammatory signal, the TNF-α/NF-κB pathway plays a central role in initiating and regulating the cascade of in ammatory factors [28,29]. In addition, TNF-α/NF-κB can also participate in the in ammatory process of diseases by inducing oxidative stress [30][31][32]. The overactivated in ammatory response after TBI releases abundant in ammatory factors, induces oxidative stress, and produces excessive oxidative free radicals including reactive oxygen species (ROS) and reactive nitrogen species (RNS). Inducible nitric oxide synthase (iNOS), a subtype of nitric oxide synthase that is only induced under pathological conditions, is regulated by NF-κB. When iNOS is activated, it can release a large amount of NO, which leads to the excessive production of ROS/RNS [33][34][35][36]. In the brain, the unchecked ROS/RNS release leads to lipid peroxidation of cell membranes. This in turn disrupts phospholipid-dependent enzymes and ionic gradients resulting in other sequelae, including disturbances of the CBF, BBB permeability changes, and the development of edema [33]. Among the many studies concerning the role(s) of ROS/RNS molecules are those that have demonstrated their major role in the signal transduction pathways associated with vascular contraction and relaxation [33,37]. Oxidative stress triggered by free radicals is closely associated with the functional regulation of vascular cells, including endothelial cells and pericytes. Free radicals, particularly RNS can impair pericyte function, stimulating pericyte contraction. Subsequently, the contracted pericytes die upon further activity of damaging factors, leading to sustained constriction of the microvasculature and affecting local blood perfusion [33,38,39].
Considering that the TNF-α/NF-κB/iNOS pathway is an important regulatory factor of the neuroin ammatory response, it may be closely related to pericyte-mediated microcirculation. Therefore, we speculated that the activation of the TNF-α/NF-κB/iNOS pathway after TBI promoted neuroin ammation and oxidative stress, which may impair cerebral microcirculation by affecting the function of pericytes.

Animals
All animal experiments were approved by the 900th Hospital Ethics Committee (Fuzhou, China) and were performed under strict supervision. Adult male C57BL/6 mice (25 ± 3g) were purchased from the experimental animal facilities of Fujian Medical University. All animals were housed at room temperature (24-26 °C), with a light-dark cycle of 10 h/14 h, and were provided with su cient water and food. Experimental model and drug administration All mice were randomly divided into three groups: a control group, a TBI group, and a TBI+in iximab (IFX) group (n = 18 each). Nine mice in each group were used for neurological assessment and cerebral blood ow monitoring, and the remaining mice were used for histological and molecular studies. The details of the damage caused by controlled cortical impact (CCI) have been described previously [40]. First, anesthesia was induced with 3% iso urane in a plexiglass container. After the animal was successfully anesthetized, anesthesia was maintained with 1.5% iso urane delivered using a smallanimal anesthetic machine (RWD Life Science Co., Shenzhen, China). Then each mouse was placed in a stereotactic frame. The craniotomy was located approximately midway between bregma and lambda on the right side, with the medial edge of the craniotomy 1 mm lateral to the midline. Mice were subjected to an impact using a 2 mm metal at-tip impactor (Brain and Spinal Cord Impactor, 68099H, RWD Life Science). The impact central point was drilled 2.5 mm to the right of the sagittal suture and 2.5 mm posterior to the coronal suture. The velocity was 5 m/s, the depth was 3 mm, and the impact duration was 100 ms. Then the scalp was closed with a suture, and the mice were put back into their home cages to recover from the anesthesia. The control group animals received identical surgical procedures without CCI. Pharmacological inhibition of TNF-α was performed using IFX, as previously described [41]. Approximately 30 min after TBI, the TBI+IFX group was intraperitoneally injected with IFX (10 μg/g, cilag Ag) once per day for 3 consecutive days. The remaining groups were injected with the same dose of the vehicle as used in the TBI+IFX group.

Cell culture
The murine microglial cell line BV2 was obtained from the China Infrastructure of Cell Line Resources (Beijing, China) and cultured in a medium comprising 90% Dulbecco's Modi ed Eagle's Medium (Invitrogen, Frederick, MD, USA), 10% fetal bovine serum (Hyclone, Logan, UT, USA), and 1% antibiotics (100 U/mL penicillin and 100 μg/mL streptomycin) at 37 ℃ in a humidi ed atmosphere of 5% CO 2 .

Preparation of para n-embedded sections
At 72 h after TBI, after deep anesthesia with sodium pentobarbital, the mice were transcardially perfused with 0.01 M phosphate buffered saline (PBS; pH 7.4) followed by 4% paraformaldehyde solution. Then, the brains were removed and post-xed by immersion in the same xative solution at 4 °C for 24-48 h.
After dehydration and vitri cation, tissue samples were embedded in para n, and 4-μm sections were prepared. The sections were then dewaxed in xylene, rehydrated in graded ethanol and deionized water, and then processed for immuno uorescence, immunohistochemistry, Nissl staining, and terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) staining.

Immunohistochemical analysis
Formaldehyde-xed specimens were embedded in para n and cut into 4-μm-thick sections that were depara nized with xylene and rehydrated in a graded series of alcohol. Antigen retrieval was carried out by microwaving in citric acid buffer. Sections were incubated with an antibody against ionized calciumbinding adapter molecule (Iba)-1 (1:500; Abcam, Cambridge, UK), ADAM17 (1:200; Boaosen Biotechnology, Beijing, China), NF-κB p65 (1:200; Santa Cruz), or iNOS (1:250; Abcam), washed and then incubated with secondary antibody for 1 h at room temperature. A total of ve sections from each animal was used for quanti cation, and the signal intensity was evaluated as follows [22]: 0, no positive cells; 1, very few positive cells; 2, moderate number of positive cells; 3, large number of positive cells; and 4, the highest number of positive cells.
Enzyme-linked immunosorbent assay (ELISA) In ammatory factors and free radicals were measured in brain tissue using ELISA kits (Jingmei Biotechnology, Jiangsu, China) for TNF-α, interleukin (IL)-1β, IL-6, interferon (IFN)-γ, ROS, RNS, and cyclic guanosine monophosphate (cGMP). According to the manufacturer's instructions, standards and samples were sequentially incubated with respective monoclonal antibodies, biotinylated anti-rat antibodies, and then horseradish peroxidase. The detected optical density (OD) values were transformed into a concentration.

Immuno uorescence staining
Formaldehyde-xed specimens were embedded in para n and cut into 4-μm-thick sections that were depara nized with xylene and rehydrated in a graded series of alcohol, followed by antigen retrieval. . After washing, the sections were incubated with secondary antibodies for 1 h at room temperature. Cell nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). Images were captured with a uorescence microscope (Leica, Wetzlar, Germany).
For the lipopolysaccharide (LPS) group, BV2 cells were stimulated with LPS (0.5 μg/mL) for 24 h, while the control group was not treated with LPS. Then the coverslips were washed with PBS three times and xed with 4% paraformaldehyde for 10 min at room temperature. Then BV2 cells were xed with 4% paraformaldehyde for 30 min, permeabilized with 0.1% Triton X-100 for 10 min, blocked with 5% bovine serum albumin (BSA) for 60 min, and incubated overnight with primary antibodies against Iba-1 (1:500; Abcam) and TNF-α (1:100; Abcam). After incubation with the secondary antibody and DAPI, images were captured with a uorescence microscope (Leica, Wetzlar, Germany).

Cytokine measurements
The supernatants were collected, and the concentrations of the cytokine TNF-α were measured using an ELISA kit (Jingmei Biotechnology) according to the manufacturer's instructions.

Nissl staining
Cortical tissue from lesioned areas was xed in formaldehyde, embedded in para n, and cut into 4-μm sections. Slices went through xylene dewaxing and an alcohol gradient rehydration as above and were stained with Nissl solution (Boster Biotech, Wuhan, China) for 5 min. Compared to normal neurons, the cell bodies of injured neurons were shrunken and/or contained vacuoles and the nuclei stained darker. A pathologist who was blinded to the experiments randomly selected ve random regions of interest (ROIs) under a high magni cation optical microscope (×400; Leica, Wetzlar, Germany) to observe positively stained cells surrounding injured areas. Five random ROIs were selected for quanti cation, and the mean (%) was used for the statistical analysis.

TUNEL staining
A TUNEL assay was performed using an apoptosis kit according to the manufacturer's instructions (Roche Inc., Indianapolis, IN, USA). Slices were incubated with NeuN (1:500; Abcam) overnight at 4 °C, and after washing in PBS, the samples were incubated with TUNEL reaction mixture for 1 h at 37 °C. TUNELpositive neurons around the injured area were observed and counted with a microscope at high magni cation (×400). Five ROIs were selected for quanti cation and averaged for statistical analysis.

Assessment of neurological injury
Nerve injury was assessed by modi ed neurological severity score (mNSS) and Garcia test [22,42]. The mNSS included motor, sensory, and re ex tests in mice. The neurological injury was recorded when a task was not completed successfully or when the corresponding re ex was lost. The mNSS test was graded on a scale of 0-18, in which a score of 0 indicated normal performance and a total score of 18 points indicated severe neurological de cits, 1-6 indicated mild injury, 7-12 indicated mean-moderate injury, and 13-18 indicated severe injury. The Garcia test consisted of seven evaluations: spontaneous activity, axial sensation, vibrissae proprioception, and limb symmetry, as well as the ability to perform lateral turning, forelimb outstretching, and climbing. Each test received a score between 0 (worst performance) and 3 (best performance), and a total Garcia score was calculated as the sum of all subtests (maximum = 21 points). The evaluation was performed pre-injury and post-injury (24,48, and 72 h) by investigators who were blinded to the experiments.
Evans blue (EB) extravasation assay BBB permeability was investigated by measuring the extravasation of EB. EB (Sigma-Aldrich; 2% in saline; 5 mL/kg) was injected via the common carotid artery 2 h prior to sacri ce 72 h after TBI [15]. Mice were transcardially perfused with ice-cold PBS (pH 7.2-7.4) immediately after sacri ce, followed by 4% paraformaldehyde in PBS. The brains were then removed, dissected, weighed, and homogenized in 600 μl 7.5% (w/v) trichloroacetic acid. The samples were then centrifuged. The absorption of the supernatant was measured using a spectrophotometer at a wavelength of 620 nm. The quantity of Evans blue was calculated according to a standard curve and expressed as micrograms of Evans blue/g of brain tissue.

Analysis of cerebral edema
The wet/dry weight method is used to evaluate brain water content, which is a reliable method for brain edema [15]. Brie y, 72 h post-TBI, brains were rapidly removed from the skull, the brain tissue was removed from the injured side with a xed weight and put in the pre-weighed aluminum foil, and samples were then placed in an oven for 72 h at 90 ℃ and reweighed for dry weight content. Brain water content was calculated according to the following formula: brain water content (%) = (total wet weight of braindry weight of brain)/total wet weight of brain × 100%.
Laser speckle contrast imaging (LSCI) Cortical blood ow was monitored using the laser speckle technique as described previously [43]. The laser speckle imaging system (Wuhan SIM Opto-technology Co., Wuhan, China) consisted of a continuous wavelength (λ = 785 nm) laser source, an Olympus ZS61 microscope, a charge-coupleddevice camera, and a computer. The selection of the ROIs in the LSCI was performed by tools provided by the software, and the values obtained were the average blood-ow values in the region. Before making the model, the regional CBF was recorded as baseline. LSCI was used to observe the relative blood ow values in the capillary areas surrounding the impinging lesion in mice before and after craniotomy, and post-injury (5 min, 24 h, 72 h). Before inducing the model, the regional CBF was recorded as the baseline.

Statistical analysis
All statistical analyses were performed using SPSS 23.0 statistical software (SPSS Inc., Chicago, IL, USA). The results are expressed as mean ± standard deviation. The comparison between two groups was performed using an independent samples t-test, while the statistical difference between each group was evaluated by one-way analysis of variance (ANOVA) with Bonferroni correction for post hoc multiple comparisons. Differences with p < 0.05 were considered statistically signi cant.

Results
Trauma induces microglial activation and promotes neuroin ammation and oxidative stress.
Microglia are a major source of in ammatory factors in the brain, and decades of studies have con rmed that microglia are the key driver of the in ammatory response in CNS diseases [6]. At 72 h after TBI, immunohistochemistry showed that the microglial marker Iba-1 increased signi cantly, suggesting that trauma induced microglial activation and proliferation (Fig. 1a). The expression levels of in ammatory factors (TNF-α, IL-1β, IL-6, IFN-γ) were measured after TBI using ELISA kits, and results showed that the TBI group had signi cantly higher expression levels of in ammatory factors compared to the control group (Fig. 1b). In addition, we found that the concentrations of ROS and RNS were also signi cantly increased after TBI (Fig. 1c). These results suggested that microglia were activated after TBI, accompanied by in ammation and oxidative stress.
Neuroin ammation and oxidative damage secondary to TBI are positively correlated with TNF-α[26]. Here, we evaluated the changes of the downstream NF-κB/iNOS pathway by immunohistochemistry. As expected, compared with the control group, TBI up-regulated the expression of NF-κB p65 and iNOS, suggesting that there was NF-κB/iNOS axis activation. ADAM17, also known as TNF-α converting enzyme, is the trigger of TNF-α pro-in ammatory activity [28]. Our results showed that the expression of ADAM17 around the injury areas in the TBI group was also increased ( Fig. 1d-g). These results further suggested that TNF-α and its signaling pathway may be closely related to the secondary injury following TBI.
Activated microglia release TNF-α and induce the activation of the NF-κB/iNOS signaling axis.
From the above experimental ndings, we found that TNF-α/NF-κB/iNOS axis activation may play an important role in the secondary injury of TBI. However, it is unclear whether this axis is associated with activated microglia, so we further explored the relationship between them in microglia (BV2 cells) cultured in vitro. After 24 h of LPS stimulation, immuno uorescence staining showed a signi cant rise in TNF-α expression levels in BV2 cells (Fig. 2a). Meanwhile, the ELISA results showed that the concentration of TNF-α in the supernatant of the culture medium also increased (Fig. 2b). Subsequently, we determined the degree of activation of the NF-κB/iNOS axis in activated BV2 cells. Western blot analyses revealed that LPS stimulation resulted in phosphorylation of NF-κB p65 and increased the expression of NF-κB p65 and iNOS (Fig. 2c). These results suggested that microglia activated the NF-κB/iNOS pathway by generating TNF-α after LPS stimulation.
Blocking TNF-α can play a neuroprotective role in the acute phase of TBI.
TNF-α is a multifunctional proin ammatory cytokine, and blocking TNF-α is neuroprotective after TBI [27]. Our results suggested that this protective effect may be associated with the inhibition of the TNF-α/NF-κB/iNOS axis. Therefore, we tested our hypothesis by inhibiting TNF-α using IFX. IFX, an anti-TNF-α agent, is a monoclonal antibody that binds to TNF-α [26]. Compared with the TBI group, the expression of TNF-α in microglia was signi cantly inhibited by IFX (Fig. 3a). Meanwhile, immuno uorescence staining and western blot analysis showed that IFX treatment could inhibit the phosphorylation and nuclear translocation of p65, and reduce the expression of iNOS (Fig. 3b-e), and this result further con rmed the association between TNF-α and the downstream NF-κB/iNOS pathway. Moreover, the ELISA results indicated that the concentrations of in ammatory factors (TNF-α, IL-1β, IL-6, and IFN-γ) and free radicals (ROS, RNS) in the TBI+IFX group were signi cantly decreased compared to the TBI group at 72 h after TBI (Fig. 3f, g). Excess NO produced by iNOS can quickly react with superoxide anions to form another powerful free radical, peroxynitrite [37,44,45]. Here, we determined the expression of 3-NT to laterally re ect peroxynitrite levels [45]. Western blot analysis showed that the expression of 3-NT was signi cantly increased after TBI, suggesting that the nitri cation of protein tyrosine residues mediated by peroxynitrite increased, and IFX treatment could down-regulate its expression (Fig. 3e).
Then, we used Nissl staining to assess the damage in neurons. At 72 h after injury, the percentage of injured cells in the TBI group was signi cantly higher than that in the control group (Fig. 4a). TUNEL staining was used to evaluate the apoptosis of neurons. The results demonstrated that TUNEL-positive neurons were increased in the TBI group compared to the control group (Fig. 4b). Western blot analyses further revealed that TBI resulted in the upregulation of apoptotic factors (cleaved caspase3, Bax) in the cortex 72 h after TBI, whereas the anti-apoptotic factor, Bcl-2, was decreased (Fig.  4c). However, these phenomena were reversed to a certain extent by IFX, indicating that inhibition of TNFα was bene cial for improving neuronal injury and apoptosis (Fig. 4a-c). We used mNSS and the Garcia test to assess neurological function in all groups. The degree of neurological de cit in the TBI group was the highest 24 h after injury and gradually recovered with time. At 72 h after injury, the neurological function in the TBI+IFX group was signi cantly improved when compared with the TBI group (Fig. 4d). These results supported our hypothesis that blocking TNF-α after TBI could play a neuroprotective role in the acute phase of injury by reducing neuroin ammation and oxidative stress, and this protective effect was associated with the inhibition of the NF-κB/iNOS pathway.
Neuroin ammation and oxidative stress secondary to TBI induce pericyte degeneration.
We aimed to understand the effect of TBI on pericytes by detecting the changes in pericyte markers. Immuno uorescence imaging analysis of the pericyte marker α-SMA showed that α-SMA and endothelial cells (CD31) were co-located in cerebral capillaries, as shown in the sections from injured and uninjured mouse cerebral cortices. At the same time, the expression of α-SMA in the TBI group decreased signi cantly compared with the control group. Interestingly, the level of α-SMA in the TBI+IFX group was signi cantly higher than that in the TBI group (Fig. 5a). Then, we analyzed the expression of α-SMA and the other pericyte markers PDGFR-β and NG2 in the lysates from mouse cerebral cortices using western blot analysis. The data showed that the expression of PDGFR-β, α-SMA, and NG2 in the TBI+IFX group was signi cantly increased compared with that in the TBI group, although it was still lower than that in the control group (Fig. 5b). Our results suggested that pericyte degeneration after TBI was associated with in ammation and oxidative stress mediated by activation of the TNF-α/NF-κB/iNOS signaling axis.
Pericyte degeneration mediates the destruction of the BBB and secondary vasogenic edema.
Pericytes are necessary for the formation of the BBB and are involved in regulating the function and structural integrity of the BBB, including the formation of TJ proteins [15]. We investigated our hypothesis that pericyte degeneration after TBI would impair the expression or formation of TJ proteins. Using triple immunostaining, we analyzed the expression of TJ proteins (occludin and ZO-1) in brain tissue slices and assessed their co-localization with endothelial cells (CD31). Our results con rmed the expression of occludin and ZO-1 were signi cantly decreased after TBI, and IFX reduced the decrease in proteins caused by trauma (Fig. 6a). Subsequently, we analyzed the expression of TJ proteins including occludin, claudin-5, and ZO-1 by western blot analysis. Similar to the results of the immuno uorescence staining, occludin, claudin-5, and ZO-1 expression levels were increased in the TBI+IFX group compared to the TBI group (Fig. 6b). Next, we further investigated the effect of the reduction in TJ proteins on BBB permeability by measuring the exudation of EB. The results showed that compared with the control group, the exudation of EB signi cantly increased in the brain tissues at 72 h after injury in the TBI group, but decreased in the TBI+IFX group (Fig. 6c).
After TBI, BBB disruption leads to increased cerebrovascular leakage, resulting in vasogenic edema. AQP4, as the most important water channel protein in the mammalian brain, is closely related to the occurrence of vasogenic brain edema [40]. The results of the western blot analysis showed that AQP4 expression was enhanced after TBI, which was inhibited by IFX (Fig. 6b). In order to further verify the consequences of BBB destruction and high AQP4 expression, we analyzed the water content of damaged brain tissue for 72 h to assess brain edema and compared this with the control group. The results showed that brain water content in the TBI+IFX group was reduced compared with that in the TBI group (Fig. 6d).

Pericyte degeneration mediates limited perfusion into the cerebral microcirculation.
Recent studies suggest that microvessels covered by pericytes play a key role in local cerebral blood ow regulation [14,46]. LSCI was used to monitor the changes in microcirculation blood ow in mice at different time points. We found that craniotomy caused minor damage to the cortical vessels, which slightly reduced the blood ow of the local microcirculation. In the TBI group, the general cerebral blood ow decreased signi cantly after impact, especially in the microcirculation around the injury areas. With the extension of time, the microcirculation around the injury areas gradually recovered in each group. In the TBI+IFX group, the microcirculation blood ow around the lesioned areas was improved at 72 h compared with that in the TBI group, which was close to the preinjury level. However, the blood ow values at other time points were not signi cantly different from those in the TBI group (Fig. 7a-c). This phenomenon is consistent with the determination of cGMP in the affected side of the brain, which is an important mediator of NO-mediated vasodilation [34,47]. The cGMP levels in the TBI+IFX group were increased at 72 h compared to the TBI group but remained lower than the control group (Fig. 7d).

Discussion
The NVU is the minimum unit necessary to maintain the proper function of CNS [10]. Cerebral microcirculation disturbance mediated by pericytes is an important factor of secondary brain injury after TBI, which is closely related to pathological processes such as neuroin ammation and oxidative stress [1,48]. Using the mouse CCI model, this study proved the importance of neuroin ammation and oxidative stress in the secondary injury following TBI. Inhibition of the in ammatory factor TNF-α can improve the prognosis of TBI. Meanwhile, we explored the effects of neuroin ammation and oxidative stress on pericyte-mediated microcirculation by assessing the TNF-α/NF-κB/iNOS pathway. We clari ed that the TNF-α/NF-κB/iNOS signaling axis in microglia affects the cerebral microcirculation by regulating pericyte function and then plays a vital role in the pathological process after TBI.
Pericytes are vascular mural cells in the NVU that play a vital role in the regulation of the microcirculation [13,14,46]. Recent studies in adult and aging brains demonstrated that pericytes are required for BBB integrity, capillary perfusion, and CBF [15]. To investigate the effects of TBI on pericytes, we used a combination of pericyte markers including α-SMA, PDGFR-β, and NG2 to visualize pericyte changes. At 72 h after TBI, the expression levels of α-SMA, PDGFR-β, and NG2 in the injured brain tissue were signi cantly decreased compared with those in the control group, suggesting the degeneration and loss of pericytes after TBI. It was also accompanied by disruption of TJ proteins (occludin, ZO-1, claudin5), which suggested damage to the BBB. Loss of brain pericyte-endothelium integrity has been shown to increase transendothelial uid ow and paracellular transport as a result of reduced TJ protein expression, both causing BBB disruption [15]. Further studies showed that the leakage of EB in brain tissue increased in the TBI group, accompanied by the increase of AQP4 expression and brain water content. These results con rmed the damage of the BBB. In order to evaluate the changes of pericytemediated microcirculation blood ow after TBI, we used LSCI to monitor cerebral cortex blood perfusion in mice. The results showed that the perfusion of the microcirculation around the injury areas was signi cantly inhibited after trauma, and gradually recovered with time.
The TNF-α/NF-κB signaling pathway is the most important pathway in the in ammatory response [49]. NF-κB is associated with the expression of numerous cytokines and is involved in the regulation of the in ammatory response, oxidative stress, apoptosis, and other pathological processes [50,51]. The NF-κB family consists of ve structurally related subunits including P50, p52, p65, RelB, and c-Rel. LPS can stimulate c-Rel activation, enhance the binding of c-Rel to the NF-κB site in the iNOS promoter and thus induce iNOS expression [35,36]. Previous studies have suggested that the activation of the TNF-α/NF-κB/iNOS axis is related to the in ammatory response and oxidative stress [45,52]. Our ndings suggested that in ammation-associated microglia were activated following trauma, accompanied by increased NF-κB phosphorylation and nuclear translocation, and upregulate iNOS expression. Further experiments in vitro also con rmed the relationship between the activation of the TNF-α/NF-κB/iNOS axis and in ammation in BV2 cells. These results are consistent with those of a previous study [53].
To verify the pathological mechanisms of TNF-α after TBI, we utilized IFX to antagonize TNF-α. IFX is a monoclonal antibody that binds to TNF-α, and its protective effects including anti-in ammation, antioxidative stress, and anti-apoptosis are thought to attenuate the injury caused by hypoxia, ischemia, edema, and vascular changes in the acute phase of TBI [26]. We found that IFX treatment inhibited the expression of TNF-α in microglia and downregulated NF-κB/iNOS pathway activity. At the same time, the expression levels of in ammatory cytokines, free radicals, and 3-NT decreased in the brain, and neuronal injury and apoptosis were reduced, as well as neurological de cits. These results further indicated the critical role of TNF-α in the secondary injury following TBI, which was associated with the activation of the NF-κB/iNOS axis.
A previous report has shown that NF-κB/iNOS pathway activation was involved in mediating pericyte apoptosis [54]. In our study, we found that inhibition of the TNF-α/NF-κB/iNOS axis by IFX could signi cantly ameliorate pericyte and TJ protein loss, thus contributing to the preservation of the integrity of BBB structure and function. We also measured the content of cGMP in brain tissue at 72 h postinjury, which is a mediator of NO-mediated vasodilation [34,47]. cGMP-related signaling is one of the pivotal mechanisms by which pericytes regulate blood ow [55,56]. Through IFX treatment, we found that microcirculatory blood ow was improved to some extent at 72 h after TBI, and cGMP levels were also increased accordingly. What is interesting is that our experimental results are contrary to Lesley's conclusion. Lesley's study based on iNOS knockout mice claims that iNOS plays a positive role in the recovery of CBF after TBI [57]. We speculate that the contradictory conclusions may be related to different means of blood ow measurement, while iNOS knockout may carry other unanticipated effects. Because NO has a complex role, both neuroprotective and neurotoxic, an appropriate concentration of NO can be of bene t in maintaining blood perfusion.
Nevertheless, this study had limitations. First, we explored the impact on pericytes after activation of the TNF-α/NF-κB/iNOS axis in microglia in the context of the NVU, but we ignored the effect of other NVU components. In addition, it remains unclear by which pathway the TNF-α/NF-κB/iNOS axis damaged pericytes, and whether pericytes underwent some speci c phenotypic transformation in this process. Therefore, future research should deeply explore the speci c mechanisms underlying the activation of the TNF-α/NF-κB/iNOS axis and mediation of pericyte degeneration, as well as the role of other components of the NVU. These questions may be answered using TNF-α or iNOS knockout mice as well as in further in vitro experiments.

Conclusions
In summary, our study demonstrated that pericyte degeneration after TBI mediated microcirculatory disturbance, which was associated with neuroin ammation and oxidative stress mediated by activation of the TNF-α/NF-κB/iNOS signaling axis in microglia. The inhibition of TNF-α affected the activity of this signaling axis, thus reducing neuroin ammation and oxidative damage, ultimately protecting cerebral microcirculatory perfusion and BBB integrity and promoting neurological functional repair. Our study proposed a possible mechanism of pericyte-mediated microcirculation dysfunction after TBI, which provided rationale for targeting pericyte as a new therapeutic target.

Availability of data and materials
The datasets analyzed during the current study are available from the corresponding author on reasonable request. TBI induces neuroin ammation and oxidative stress. a The expression of the microglial marker, Iba-1, increased signi cantly at 72 h after TBI. Scale bars = 20 μm. b, c TBI promotes the production of in ammatory factors (TNF-α, IL-1β, IL-6, and IFN-γ) and free radicals (ROS, RNS) around the injury areas. d, e After trauma, the expression of NF-κB p65, iNOS, and ADAM17 increased in the lesioned area.

Figure 2
Activated microglia upregulate the expression of TNF-α, which promotes NF-κB/iNOS signaling in vitro. a At 24 h after LPS stimulation, BV2 cells (Iba-1+) had increased TNF-α expression. b ELISA showed that the concentration of TNF-α in the culture medium increased signi cantly after LPS stimulation. c Activation of NF-κB p65 and iNOS in BV2 cells treated with LPS for 24 h. Values are expressed as the mean ± standard deviation (n = 6 per group), *p < 0.05, **p < 0.01. Scale bars = 50 μm IFX reduces neuroin ammation and oxidative damage by inhibiting the activation of the TNF-α/NF-κB/iNOS pathway. a IFX inhibited the expression of TNF-α in microglia (Iba-1+). b Immuno uorescence staining showed that IFX treatment signi cantly inhibited the nuclear translocation of NF-κB p65 in microglia (Iba-1+). c TBI enhanced the expression of iNOS, which was decreased by IFX supplementation.
Representative photomicrographs of iNOS are shown. d IFX supplementation signi cantly decreased the Page 22/26 expression and phosphorylation of NF-κB p65 and inhibited the translocation of NF-κB p65 from the cytoplasm to the nucleus. e Western blot analyses revealed that TBI resulted in the upregulation of iNOS and 3-NT in the cortex; however, compared with the TBI group, the levels of both were decreased in the TBI+IFX group. f, g ELISA results showed that inhibition of TNF-α could signi cantly decrease TBIinduced enhancement of in ammatory factors and free radicals. Values are expressed as the mean ± standard deviation (n = 6 per group), *p < 0.05, **p < 0.01. Scale bars = 50 μm Figure 4 Inhibition of TNF-α can play a neuroprotective role after TBI. a The percentage of damaged cells was higher in the TBI group than in the control group; the damaged fraction was signi cantly lower in the TBI+IFX group than in the TBI group. Representative photomicrographs of Nissl-stained neurons are shown. Scale bars = 50 μm. b IFX supplementation signi cantly decreased the rate of TUNEL-positive neurons after TBI. Scale bars = 20 μm. c Western blot demonstrated that the expression of apoptotic factors (cleaved caspase3 and Bax) was signi cantly reduced in the TBI+IFX group compared with the TBI group, along with an increase in the anti-apoptotic factor (Bcl-2). d Antagonizing TNF-α ameliorates neurological de cits 72 h after TBI. Values are expressed as the mean ± standard deviation (n = 6 per group), *p < 0.05, **p < 0.01. Figure 5 TBI results in loss of pericyte coverage. a Immuno uorescent staining of α-SMA merged with endothelial cells (CD31+) in mouse brain cortex tissue samples of the control, TBI, and TBI+IFX groups at 72 h after injury. b Western blot analysis of α-SMA, PDGFR-β, and NG2 in mouse cortical tissue lysates from the control, TBI, and TBI+IFX injury groups 72 h after TBI. Bar diagram represents the results expressed as a ratio of pericyte markers and β-actin bands. Values are expressed as the mean ± standard deviation (n = 6 per group), *p < 0.05, **p < 0.01. Scale bars = 20 μm. TBI destroys the BBB and aggravates brain edema by mediating pericyte degeneration. a Representative images of uorescent staining of occludin and ZO-1 merged with endothelial cells (CD31+) in mouse brain cortex tissue samples of the control, TBI, and TBI+IFX groups at 72 h after injury. b TJ proteins (occludin, ZO-1, claudin5) and AQP4 in different groups measured by western blot. c The TBI group had more EB extravasation in the brain 72 h after TBI. Compared with the TBI group, the TBI+IFX group had signi cantly decreased EB extravasation. Representative photos of EB extravasation in the experimental groups. d Antagonizing TNF-α decreased brain water content 72 h after TBI. Values are expressed as the mean ± standard deviation (n = 6 per group), *p < 0.05, **p < 0.01. Scale bars = 20 μm. Pericyte dysfunction affects microcirculatory blood ow after TBI. a Craniotomy slightly damaged the cortical vessels, while CCI caused severe damage to the vessels in the injured area. The red circle in the gure is the preset strike zone. Scale bars = 0.5 mm. b Representative LSCI blood-ow map at different time points in the control, TBI, and TBI+IFX groups. The white circle marked in the gure is the region of interest (ROI). c After impact, the blood ow in the microcirculation area decreased signi cantly in the TBI group, and then gradually recovered. Inhibition of TNF-α could improve local blood ow 72 h after impact, but there was no signi cant difference at other time points. d The ELISA results showed that the content of cGMP in the TBI+IFX group was signi cantly higher than that in the TBI group. Values are expressed as the mean ± standard deviation (n = 6 per group), *p < 0.05, **p < 0.01.