Background
Spinal cord ischemia/reperfusion (I/R) injury is the most devastating complications encountered in many pathophysiological situations, such as hypotension, surgical procedures on thoracic, thoracoabdominal aneurysms and the spine [
1]. It remains as a widespread and persistent problem, because its debilitating injuries to the central nervous system results in high incidence of paraplegia posing a serious threat to patients. However, the underlie mechanism of spinal cord I/R injury is not well understood. This is probably due to the induction of spinal cord I/R injury is multifactorial [
2‐
4]. Blood–spinal cord barrier (BSCB), surrounded by astrocytes and perivascular microglia, consists of a continuous capillary endothelium with tight junctions between the cells. As shown in our previous studies, BSCB disruption and inflammatory reactions play an important role in the evolution of spinal cord I/R injury and in promotion of neuronal damage [
5,
6]. Research has expanded into the glial/neuronal transmission and immune responses of resident glial cells to spinal cord injury [
7,
8]. Existing evidence shows that spinal glial activation involves important components of the immune system and triggers rapid signal transduction cascades of the transcriptionfactor nuclear factor κB (NF-κB), driving gene expression of proinflammatory cytokines (e.g. IL-1β) in the course of pathophysiological changes that occur after brain injury [
9]. Nonetheless, the specific cellular source within microglia which is responsible for transferring immune stimuli into the nervous system responses is unknown. During the pathogenic cascade after central nervous system (CNS) injury, microglia are thought to be the first nonneuronal cells to express a plethora of growth factors, chemokines, and regulatory cytokines as well as free radicals and other toxic mediators [
7,
10].
In microglia, Toll-like receptors (TLRs), especially TLR
4, have been shown to recognize various microbial products and to initiate innate immune responses upon interaction with infectious agents or endogenous ligands present in the spinal cord
in vivo and
in vitro[
10,
11]. The general understanding is that TLR
4 can be specifically activated to initiate immune responses by presenting a pathogen-derived antigen to naïve T cells when sensing lipopolysaccharide (LPS), a common constituent in the cell wall of Gram-negative bacteria [
12].Studies have demonstrated that TLR
4 is upregulated in different I/R-injured organs, including heart, brain, liver, and kidney [
13,
14]. Nevertheless, the relationship between TLR
4 and microglial activation in spinal cord I/R injury remains unknown. Thus, we hypothesized that there might be a link between TLR
4 of microglia to the stressors that damaged neurons: endogenous antibodies or cytokines that leaked through disrupted BSCB, which, in turn, contributes to further activation of microglia and BSCB disruption. Moreover, recent studies suggest that microglia activation which could be inhibited by minocycline is a double-edged sword in various neurological models and under different conditions [
8,
15]. Ischemia was regarded as a powerful stimulus that disabled the endogenous inhibitory signaling and triggered microglial activation [
7]. Upon activation, microglia could exhibit plenty of phenotypes and release both pro- and anti-inflammatory mediators to either exacerbate ischemic injury or help repair. A few studies, however, have conducted on the spinal cord model of I/R injury and state that whether activation of microglia is deleterious and/or beneficial for spinal cord recovery is still a controversial topic. To testify our hypothesis, we first explore whether microglia are activated in a rat model of spinal cord I/R injury. Next, we evaluate the roles of TLR
4 and NF-κB during I/R-induced BSCB disruption with and without the involvement of microglia.
Discussion
Ischemia/reperfusion (I/R) injury of the spinal cord after an operation on the thoracic aorta is an unpredictable, however, disastrous complication. To the best of our knowledge, the current study is the first to demonstrate activation of TLR4 in microglia and we show TLR4 is participates in inflammatory reactions in the blood–spinal cord barrier (BSCB) after I/R injury. Early and sustained microglial activation can be determined by the mRNA and protein expression of the microglial surface markers, Iba-1. The membrane-bound receptor TLR4 is also overexpressed in the spinal cord but not in sham-operated rats from 12 to 36 h after I/R injury. In addition, intrathecal administration of minocycline for 3 days before ischemia showed obvious protective effects in the form of a decrease in I/R-induced microglial activation, in BSCB disruptions, and in the production of proinflammatory cytokines. Results from this study also demonstrated that when administered in this preemptive manner, intrathecal infusion of a TLR4 receptor inhibitor, TAK-242, exhibits protective effects similar to minocycline pretreatment results, whereas intrathecal injection of LPS, a TLR4 agonist, exacerbates deleterious effects. Given that NF-κB is known to be activated in the presence of TLR4, it is expected that the NF-κB inhibitor,PDTC inhibit I/R-induced microglial activation via upregulating TLR4.
The I/R model used and the period observed were referred to previous studies, in which the evolution of inflammatory cytokine activation in the spinal cord peaks between 24 and 48 h after reperfusion [
5,
6,
16]. This model is a reliable and stable animal model for studying neuroprotective manipulations and molecular mechanisms in the spinal cord. BSCB consists of a continuous capillary endothelium with tight junctions between its cells, surrounded by astrocytes and perivascular microglia. The intact barrier can prevent vasogenic edema and pathological effects on CNS by restricting access of molecules and cells to the spinal cord under conditions of stroke and I/R [
17,
18]. It is generally believed that inflammatory factors play a critical role in leakage of BSCB as a result of aberrant vascular permeability due to dissociation of zonula occludens-1 (ZO-1) from the cytoskeletal complex and due to an increased level of matrix metalloproteinases (MMPs) and tumor necrosis factor α (TNF-α) [
5,
6,
17‐
19]. As demonstrated in our previous study, BSCB permeability is altered in the course of spinal cord I/R injury, and these changes could be measured using extravasation of EB dye [
5,
6]. This damage may caused by inflammatory processes and it may further exacerbate inflammation. Recent evidence support the notion that microglia, components of the immune system are activated and play an important role in the initiation phase of I/R-induced neurodegenerative and inflammatory processes. When triggered by peripheral inflammation or nerve injury, spinal microglia can be rapidly activated and can respond to the neurotransmitters released by central terminals of primary sensory neurons, such as glutamate, substance P, and adenosine triphosphate. After that, the release from activated glial cells of a series of growth factors, chemokines, regulatory cytokines as well as free radicals and other toxic mediators such as IL-1β, TNF-α, prostaglandin E
2, and reactive oxygen species (ROS), results in activation of rapid signal transduction cascades leading to either survival or death of neurons [
9,
17]. Representative micrographs of ionized calcium–binding adaptor molecule 1 (Iba-1) are commonly used to quantify activated microglia. Upon activation, microglial cells transform from the ramified shape to rounded (amoeboid) macrophage-like morphology [
20]. There are significantly greater numbers of Iba-1–positive cells in the I/R group both at 12 and 36 h in comparison to the lower numbers in the sham group. Furthermore, the colocalization with TLR
4 according to double-immunofluorescence analysis confirmed that TLR
4 is indeed upregulated in activated microglial cells in injured regions of the spinal cord.
There are substantial protective effects of minocycline (a member of the tetracycline antibiotic family), which prevent microglial activation and generation of glutamate, IL-1β, and nitric oxide (NO) [
21]. To explore the role of microglia in BSCB disruption after I/R injury, minocycline was infused intrathecally during 3 days before the surgical procedure in the present study. We also found that inhibition of microglial activation by minocycline is accompanied by a decreased number of Iba-1–positive cells.
Blamire et al. [
22] and van Vliet et al. [
23] examined the role of proinflammatory cytokines in BSCB leakage. These investigators reported that increased plasma levels of inflammatory cytokines (such as IL-1β and IL-6) in residual microglia (where disruption of the blood–brain barrier occurred) might also be responsible for BSCB leakage. At the same time, we observed another protective role of minocycline: it seems to attenuate (1} the BSCB disruption as measured by EB extravasation and (2} an increase in spinal water content. Our data show that minocycline decreases the number of Iba-1–positive cells at 12 and 36 h after reperfusion and attenuates upregulation of the proinflammatory molecules IL-1β and NF-κB in the spinal cord. These findings were corroborated by previous studies not only in experiments with ischemia [
3,
4] but also in studies of systemic proinflammatory states [
9,
23,
24]. These observations suggest that activated microglia may potentiate damage to BSCB components, which is caused at least in part by proinflammatory cytokines.
Previously, we elucidated the molecular mechanisms underlying inflammatory and immunological processes of microglial activation after I/R. Upon activation, microglial cells start to express TLR
2–4 on their surface [
11,
14]. Among diverse TLRs, it has been reported that TLR
4 enables microglia to induce immune and inflammatory responses and to release massive amounts of proinflammatory cytokines via activation of NF-κB, once TLR
4 binds to its endogenous or exogenous ligands. Studies show that TLR
4-deficient mice display significantly attenuated behavioral hypersensitivity and are characterized by weaker spinal glial activation and lowered release of proinflammatory cytokines [
8,
12,
14]. There is usually no viral or bacterial infection in I/R models; recent research supports the notion that upregulated expression of TLR
4 is implicated in activation of microglia in cerebral ischemia models [
8,
16]. Nonetheless, there is still debate and controversy regarding the role of microglial TLR
4 in the spinal cord during the earliest stage (<3 days) after I/R, where TLR
4 must be involved in various signal transduction pathways. Our results show that early, robust, and sustained microglial activation after I/R injury is characterized by a marked long-term induction of the TLR
4 expression at protein and mRNA levels, matching the pattern of immunostaining with Iba-1. In parallel, cytokines IL-1β are released in the spinal cord between 12 and 36 h after reperfusion, the finding that is consistent with the above reports. We also confirmed the specificity and the role of TLR
4 in the microglial activation and in inflammatory reactions induced by intrathecal infusion of TLR
4 receptor antagonists or agonists. We found that downregulation of TLR
4 receptor by TAK-242 lowers the amount of resident microglial activated cells, decreases levels of NF-κB translocation, and consequently downregulates the cytokine IL-1β in the spinal cord between 12 and 36 h after reperfusion. Because TLR
4 is a known specific receptor for Gram-negative bacterial components (LPS), the rats treated with LPS showed aggravated inflammatory processes, and BSCB disruption occurred in the I/R region, in keeping with the pattern of upregulated TLR
4.
In view of the pivotal role of TLR
4 in microglial activation in the initial phase of I/R injury, the question arises how TLR
4 mediates microglial activation in spinal cord I/R injury. After activation of the TLR
4 signaling pathway, NF-κB relocates to the nucleus and regulates expression of target inflammatory genes via modulation of both MyD
88 or MyD
88 adaptor–like adaptor protein and TIR domain–containing adaptor [
12,
16,
25]. With double immunofluorescence in present study, we provided evidence that NF-κB signaling pathway activated by TLR
4 receptor played important roles in regulating immune and inflammatory responses after I/R. We detected greatly upregulation of NF-κB and IL-1β (at the mRNA and protein level) in spinal cords limited to the ischemic region between 12 and 36 h after I/R injury and was accompanied with a selective increase in TLR
4 expression. Pyrrolidine dithiocarbamate (PDTC) is a low-molecular-weight thiol compound, was initially regarded as a potent inhibitor of NF-κB by inhibiting factor I-κB phosphorylating, thus preventing the dissociation of the NF-κB -IκB complex and interfering with the generation of proinflammatory cytokines [
26‐
28]. It has been reported that intrathecal PDTC can delay and reverse mechanical allodynia in several neuropathic pain conditions [
27]. Similarly, in present study, intrathecal infusion of PDTC was observed to have suppressive effects on mechanical allodynia, BSCB dysfunction and microglial activation as well as up-regulated TLR
4 mRNA and protein expressions in spinal cord, suggesting that the activity of NF-κB pathway could regulate spinal cord I/R injury via regulating TLR
4.
Neurons play very important roles in the nervous system, involved in the processes of memory, sense and behavior. A result of I/R-induced neurological deficit was partly contributed to apoptosis of neurons in spinal cord as reported in our previous studies [
2,
5,
6]. Our data of NeuN immunoreactivity sufficiently evidenced I/R lead to a decrease neuron number in ventral gray matter and increase percentage of double-labeled cells with cleaved capase3, which referred as a mark of apoptosis. Meanwhile, microglia may release various factors to support and guide the transfer of neurons, participate in neurons repair and regeneration [
29,
30]. IL-1β is one of the final inflammatory molecules produced in TLR
4 signaling pathway, which elicits a cascade of activation of cytokines and multiple biological effects [
22]. In a variety of inflammatory conditions, reasonable upregulation of IL-1β has been shown to limit extreme inflammatory responses, with a possibility of rapidly activated phagocytosis of dead or dying cells to prevent a release of a cascade of proinflammatory cytokines and to resolve inflammation. Nevertheless, excess IL-1β can significantly worsen inflammation and tissue injury [
9,
13,
22]. Thus, these signals should be regulated very tightly to balance proinflammatory and anti-inflammatory pathways. Brikos and colleagues found that the cytoplasmic portion of TLRs, called the Toll/IL-1 receptor (TIR) domain is highly similar to that of the IL-1 receptor family [
1]. Our study confirmed that IL-1β expression and loss and apoptosis of neurons triggered by TLR4/ NF-κB signal in spinal cord were strongly increased by intrathecal infusion of LPS; in contrast, the expression levels were much lower in rats treated with minocycline, TAK-242 or PDTC, indicating that there might be a positive feedback circuit. In other words, the expression of inflammatory cytokines is regulated by TLR
4 in microglia, and in turn, inflammatory cytokines can cause microglial activation by amplifying and maintaining inflammatory response via the TLR
4 pathway. This notion also offers a probable reason for the difficulty in developing an effective therapy for spinal cord injury–related complications. Interestingly, the similar findings have been reported by Bell at al in a mice model of aortic cross-clamping, recently [
31]. That study showed that inhibition of TLR
4-mediated microglial activation may be a major mechanism of neuroprotection associated with the anti-inflammatory effects. Regarding the controversial effects of microglial activation via its plenty of membrane-bound receptors for spinal cord recovery under different stimuli, further studies would be required to clarify the complicated role of microglial TLR
4 signaling in models of I/R injury. As potential therapeutic modalities, minocycline, TAK-242 and PDTC would need further research, including safety testing; for example, their effects on learning and memory are unknown.
Based on our results, we conclude that inhibition of microglial activation and proliferation via the TLR4–microglia–NF-κB/IL-1β pathway result in protective effects in some neurological deficits, hence, fortifying BSCB integrity, and reducing spinal cord swelling.
Materials and methods
Experimental animals
All experimental procedures were approved by the Ethics Committee of China Medical University and were in compliance with the Guide for the Care and Use of Laboratory Animals (U.S. National Institutes of Health publication No. 85–23, National Academy Press, Washington DC, revised 1996). Male Sprague–Dawley rats, weighting 200–250 g were used in this study. All rats were maintained under standard condition throughout the experimental period. Either motor or sensory dysfunction was observed in rats that intrathecally received minocycline, lipopolysaccharide (LPS), TLR4 inhibitor (TAK-242), pyrrolidine dithiocarbamate (PDTC) or saline before the induction of ischemia.
Experimental I/R spinal cord injury
The spinal cord I/R model was induced by occlusion of the aortic arch for 14 minutes, as previously reported [
16].All rats were anaesthetized with intraperitoneal injection of 4% sodium pentobarbital at an initial dose of 50 mg/kg. Lung ventilation was achieved using a mouse ventilator (Hollinston, MA) with endotracheal intubation. Body temperature was continuously monitored with a rectal probe and was maintained at 37.5 ± 0.5°C with the aid of a heated operating table. A catheter was inserted into the left carotid artery and into the tail artery to measure proximal and distal blood pressure (Spacelabs Medical Inc., Redmond, WA, USA). Under direct visualization, the aortic arch was cross-clamped between the left common carotid artery and the left subclavian artery. A catheter was inserted into the left carotid artery and into the tail artery to measure proximal and distal blood pressure. Ischemia was confirmed as a 90% decrease in flow measured at the tail artery by a laser Doppler blood flow monitor (Moor Instruments, Axminster, Devon, UK) for 14 min, after which, the clamp was removed and 36 h of reperfusion took place. Sham operation rats underwent the same procedure, but no occlusion of the aortic arch was performed.
The experimental protocol
One hundred twenty rats were randomly assigned to 1 of 6 groups by means of a random number table: the I/R group (n = 24), I/R + minocycline (I/R + M) group (n = 24), I/R + lipopolysaccharide (I/R + L) group (n = 24), I/R + TLR4 inhibitor (TAK-242, I/R + T) group (n = 24), I/R + pyrrolidine dithiocarbamate (PDTC, I/R + P) group (n = 24) or the sham group (n = 24). Spinal cord I/R injury was induced by occlusion of the aortic arch for 14 min, whereas the aorta was exposed, but without occlusion in the sham group. In all groups, we performed intrathecal infusion of 10 μL normal saline, 10 nmol/μL minocycline (10 μL, Nichiiko, Toyama, Japan), 1 nmol/μL LPS (10 μL, Sigma; E. coli 011:B4), 10 nmol/μL TAK-242 (10 μL, EMD, Millipore; CAS 243984-11-4), 100 pmol/μL PDTC (10 μL, Sigma-Aldrich, Mo;71935) or 10 μL of saline respectively, continuously for 3 days before the surgical operation. The rats were euthanized 12 and 36 h after the surgical procedure. At each time point, animals were anesthetized with an overdose of pentobarbital and the L4–6 segments of spinal cords were rapidly collected for analysis because of their vulnerability to ischemic injury.
Behavioral analysis
To quantify mechanical allodynia, the withdrawal threshold of a hind-limb paw was assessed using von Frey filaments (Stoelting Co., Wood Dale, IL, USA) and the Dixon up–down method as described by Chaplan and coworkers [
32]. All behavioral tests were performed before the surgical procedure (baseline) and at 12-h intervals during a 36-h observation period by an observer who was blinded to the experimental procedures.
Measurement of spinal cord edema
Water content of the spinal cord was measured by means of the wet–dry method as quantitative measurement of edema, as reported previously [
5,
6]. The percent water content was calculated using the following formula: %H
2O = (wet weight − dry weight) × 100/wet weight.
Measurement of Evans blue extravasation
After survival of rats for 12 and 36 h, Evans blue (EB) content and EB fluorescence were used for quantitative and qualitative analysis of BSCB disruption after spinal cord I/R injury, as described previously [
5,
6]. Briefly, EB at 30 g/L (45 mg/kg; Sigma) was slowly intravenously injected into the tail vein 60 min before euthanasia. After being adequately perfused with saline under deep anesthesia, the L
4–6 segment was removed and soaked in methanamide for 24 h (60°C) and then centrifuged. EB content was measured as absorbance of the supernatant at 632 nm on a microplate reader (BioTek, Winooski, VT) and calculated as the amount of EB per wet tissue weight (μg/g). For measurement of the fluorescence, the tissue was fixed in 4% paraformaldehyde, sectioned (10 μm), and kept frozen and sealed in a light-tight container. EB staining was visualized using a BX-60 (Olympus, Melville, NY) fluorescence microscope (green filter). Percentage of recognized area (fluorescence intensity above the threshold) referred to the whole image area was performed using Image J software (NIH Image, Bethesda, MD).
Iba-1 immunoreactivity
Microglia were stained using an antibody against the microglial marker, ionized calcium–binding adaptor molecule 1 (Iba-1) as described previously [
17,
32]. Briefly, the sections were firstly blocked with 10% bovine serum albumin for 1 h at room temperature. After that, the sections were incubated with a primary rabbit anti–Iba-1 antibody (1:800: Wako, 019–19741) at 4°C overnight. After incubation with an Alexa 488–conjugated donkey anti–rabbit IgG antibody (1:500; Molecular Probes, Rockford, USA) for 1 h, the stained sections were examined under a microscope (Carl Zeiss Axio Observer Z1, Jena, Germany) determined the number of immunoreactive cells in the medial superficial dorsal horn (laminae I–III). Nonspecific staining was determined by omitting the primary antibody. The data were calculated as average numbers of positive cells per area of a spinal section ± standard error of the mean (SEM).
Double immunofluorescence
Double immunofluorescence analysis was carried out to confirm the expression of TLR4 in microglia and explored the relationship with NF-κB signal pathway and neuroapoptosis after I/R [
5,
32]. Briefly, spinal cord was fixed and sectioned into 10-μm slices with a Leica CM3050 S cryostat. The sections were blocked with 10% bovine serum albumin (BSA) for 1 h at room temperature and incubated overnight at 4°C with the primary antibodies: mouse anti-TLR4 (1:100, Abcam), mouse anti-cleaved caspase3 (1:400, Cell signal technology), rabbit anti-Iba-1 antibody (1:800, Wako), rabbit anti-NF-κB p65(1:500, Abcam), rabbit anti-NeuN (1:800, Abcam). After incubation with Alexa 594-conjugated donkey anti-mouse IgG (1:500, Molecular Probes) and Alexa 488-conjugated donkey anti-rabbit IgG (1:500, Molecular Probes) for 2 h at room temperature. Each of the steps above was followed by four rinses 5–10 min each in PBS containing 10% BSA and 0.25% Triton X-100. Images were captured using a Leica TCS SP2 (Leica Microsystems, Buffalo Grove, IL, USA) laser scanning microscope and photographed by the attached digital camera to determine the number of immunoreactive cells. Nonspecific staining was determined by omitting the primary antibody. The data were expressed as numbers of positive cells/area/spinal section ± standard error mean (SEM).
Measurement of IL-1β using ELISA
The spinal cord was collected and homogenized, followed by centrifugation. The IL-1β content was determined using an ELISA kit (R&D Systems, Minneapolis, MN, US). According to the manufacturer’s instructions, absorbance (A) was quantified at λ = 450 nm. The IL-1β content of each sample was calculated based on the standard curve, and IL-1β concentration was expressed in pg/mg protein.
Western blots
The protein expression of TLR4 and NF-κB p65 in spinal cord tissue was determined using Western blotting analysis. The rats’ spinal cords were homogenized and nuclear and cytoplasmic extracts was purified from each specimen by using Nucleoprotein and cytoplasmic protein extraction kit according to the manufacturer’s instructions (KGP-150; KangChen, Shanghai, China). The antibodies used in this experiment were mouse monoclonal anti-TLR4 (Abcam), rabbit polyclonal anti–NF-κB p65 (phospho S536, Abcam), mouse monoclonal anti–Histone (Abcam) and anti-mouse GAPDH (dilution 1:10,000, Abcam) overnight on a shaker at 4°C.After three washes with TBS-0.1% Tween, the membranes were incubated with horseradish peroxidase-conjugated secondary antibodies(Bioss, Beijing, China) for 1 h. Semiquantitation of scanned images was performed using Quantity One software (Bio-Rad Laboratories, Milan, Italy).
Real-time PCR
Quantitative real-time PCR was performed as described previously [
20,
21]. Total RNA was extracted from L
4–6 spinal cord tissue using the TRIzol reagent (Invitrogen–Life Technologies), following the manufacturer’s instructions. PCR was performed as described previously using a SYBR Green SuperMix-UDG and was conducted on a Prism 7000 detection system (Applied Biosystems, Foster City, CA). The following primers were used; TLR
4 (NM_0191178, 127 bp): forward 5′-GGATGATGCCTCTCTTGCAT-3′, reverse 5′-TGATCCATGCATTGGTAGGTAA-3′; NF-κB (HL26267H): forward 5′-CTTCTCGGAGTCCCTCACTG-3′, reverse 5′-CCAATAGCAGCTGGAAAAGC-3′; and
GAPDH (HNM_023964H, 238 bp): forward 5′-AGAAGGCTGGGGCTCATTTG-3′, reverse 5′-AGGGGCCATCCACAGTCTTC-3′. Amplification was performed using the following cycling conditions: 50°C for 2 min (uracil-DNA glycosylase incubation), 95°C for 10 min, followed by 40 cycles of denaturing at 95°C for 15 seconds and annealing at 60°C for 30 seconds. All reactions were performed in triplicate. Gene expression was calculated relative to the endogenous control samples (
GAPGH) to obtain a relative quantity (RQ) value (2-ΔΔCt, where CT is the threshold cycle).
Statistical analysis
All data were collected by investigators blinded to surgery status of the rats. The data were calculated as mean ± SEM and analyzed using the SPSS software (version 17.0, SPSS Inc., Chicago, IL, USA). The statistical data were processed with one-way analysis of variance (ANOVA) followed by Newman–Keuls
post hoc analysis. Differences with a
P value of <0.05 were considered statistically significant (Additional file
1: Figure S1).
Authors’ contributions
X-QL and BF participated in the animals’ care and made all the animal models. X-QL, JW and W-FT participated in tissue preparation, and sectioning and performed most immunohistochemistry; X-QL, BF and W-FT performed western blotting assay and statistical analysis; HM involved in the guide of model design and study design; JW gave important directions to data analysis and manuscript writing. All authors read and approved the final manuscript.