Sevoflurane preconditioning ameliorates neuronal deficits by inhibiting microglial MMP-9 expression after spinal cord ischemia/reperfusion in rats

Spinal cord ischemia/reperfusion (IR) injury is a well-recognized clinical problem resulting in motor and sensory dysfunction that is characterized as temporary or permanent ischemia of the spinal cord following reperfusion in a setting of shock or thoracoabdominal aorta surgery [1]. The blood brain barrier (BBB) is a well-known specialized structure that maintains brain tissue in an immune-privileged environment [1]. The spinal cord, as a part of the central nervous system (CNS), connects the brain and the peripheral nervous system. The blood-spinal cord barrier (BSCB) was presumed to represent both an anatomical and functional unit mediating molecular transport and immune regulation. The BSCB, consisting of a layer of tightly adhering endothelial cells lining the blood vessel lumen surrounded by astrocytes, pericytes and perivascular microglia, is critical for the barrier function by actively and selectively restricting the passage of water, ions, metabolites, and immune cells into the spinal cord under physiological and pathological states [2]. In a previous study using a rat IR model, we documented that increases in BSCB dysfunction and inflammatory reactions play an important role in the evolution of spinal cord IR injury and in the progression of neuronal damage [3],[4]. Furthermore, accumulating evidence suggests that the extracellular matrix and tight junction complexes serve several important physiological functions, including defense and homeostatic balance between the spinal cord and systemic circulation, aberrant decomposition of which could cause neurotoxic effects [5],[6]. Treatments that prevent decomposition of the extracellular matrix and tight junction complexes after IR may improve the prognosis of patients with spinal cord IR injury.

Matrix metalloproteinases (MMPs), a family of proteolytic enzymes, influence tissue injury and repair [7],[8]. Among multiple nociception-related MMPs, levels of MMP-9, which is mainly secreted by glial cells, were markedly increased in cerebral IR models [7]. During the injury procedure, ischemia-induced oxidative stress and cytokines increased MMP-9 levels, which ultimately remodeled microvascular structures of the BBB and led to an influx of immune or inflammatory cells as well as subsequent secretion of chemokines and cytokines, thereby additionally stimulating both proliferation and recruitment of inflammatory cells into the CNS. Therefore, this self-sustaining vicious cycle of inflammatory damage to the BBB is tightly regulated by the expression of MMPs [8],[9]. However, whether or how MMP-9 responds during spinal cord IR injury remains unknown.

Microglia, the primary immune effector cells of the spinal cord represent an important source of MMPs in the CNS [9]. They are situated very close to the BSCB to facilitate their robust activation in response to insult through a highly regulated network of cytokines and chemokines to subsequently orchestrate immune responses at sites of injury following BSCB opening [3],[4],[9],[10]. Existing evidence shows that these extracellular matrix proteins could be a strong inducer of microglial activation [11], switching microglia from a resting state into an activated potentially destructive phenotype characterized by morphological changes, migration, proliferation, phagocytosis, antigen presentation, and secretion of diffusible factors including proinflammatory cytokines (such as TNF-α and IL-1β) [4],[10],[11]. Moderate activation of microglia is necessary for host defense, whereas over activation of microglia, together with excess secretion of proinflammatory chemokines and cytokines, is neurotoxic and promotes neuronal injury [3],[10]. Recent papers report that neuronal cell death causes microglial activation by releasing various signaling molecules such as the chemokines, gamma interferon-inducible protein (CXCL) 10 and CC chemokine ligand (CCL) 2, as well as proinflammatory cytokines (e.g. IL-1β) and active forms of MMPs [9],[12],[13]. Thus, modulation of microglial activation and MMP-9 expression is considered to be a promising therapeutic strategy for IR injury. One control mechanism of MMPs is the interaction with TIMPs (tissue inhibitors of MMPs) [14]. Imbalances between MMPs and their inhibitors, resulting in excessive MMP proteolytic activity, are implicated in the pathogenesis of neuroinflammation [9],[14]. Generally, all MMPs are inhibited by TIMPs once activated [14],[15]. Therefore, we modulated MMP-9 expression by intrathecal injection of specific siRNAs targeting MMP-9 to further explore if IR-induced dysfunction is due to microglial MMP-9 [15].

Sevoflurane is the most widely used volatile anesthetic for both induction and maintenance of anesthesia in clinical practice. Mounting evidence indicates that sevoflurane preconditioning might induce ischemic tolerance in the brain and heart, and alleviate hypoxic and ischemic injury [16]-[18]. These findings have triggered an increasing interest in both basic science and clinical research. Previous studies demonstrate that sevoflurane can protect against ischemic neuronal damage and attenuate necrosis and apoptosis through its capacity to increase cerebral blood flow and preserve cerebral autoregulation [19]. However, few studies have attempted to describe the protective effect of this agent on spinal cord IR injury [16],[20],[21]. Therefore, the primary objective of this study was to investigate whether sevoflurane preconditioning can initiate protective effects on hind limb motor function and improve the viability of motor neurons in a rat model of transient aortic arch occlusion. Secondary objectives were to evaluate the correlation of microglial MMP-9 levels with downstream proinflammatory and apoptotic cytokines, such as CXCL10, CCL2, IL-1β, and caspase-3, with or without sevoflurane pretreatment.

All rats had normal neurological signs before the induction of ischemia and maintained good health throughout the experimental period.

Physical damage to the spinal cord after ischemia/reperfusion (IR) injury is accompanied by increased blood spinal cord barrier (BSCB) permeability, followed by neuronal apoptosis, glial activation, and prolonged neuroinflammatory responses [6],[14],[25]. Considerable efforts have been directed at increasing ischemic tolerance in the spinal cord in both experimental animals and humans in order to decrease the incidence of paraplegia. Unfortunately, current therapies remain unsatisfactory and surgical management is still the main choice of treatment [4],[6],[26]. Sevoflurane, a volatile anesthetic with the unique clinical characteristics of rapid recovery time with little emergence agitation, is suitable for both induction and maintenance of anesthesia and as a result is the most frequently used anesthetic, clinically. Neuroprotective effects of sevoflurane preconditioning have been reported in several models of cerebral ischemia and are attributed to its capacity to increase cerebral blood flow for autoregulation [16],[18],[19]. Given the benefits of volatile anesthetic preconditioning in improving myocardial functional recovery and reduced infarct size after ischemia [17],[27], we explored the therapeutic potential of treatment with sevoflurane before ischemic injury to induce a "preconditioned state" upon IR injury. Consistent with previous research in our laboratory [16],[19], the present study demonstrated that spinal cord IR injury induced by 14-min aortic arch occlusion produced extensive neuronal apoptosis manifesting as severe hind-limb motor functional deficits, along with increased BSCB leakage. Sevoflurane preconditioning 1 h prior to ischemia led to significant improvements in BSCB integrity and motor function due to decreased microglial activation resulting in reduced expression of MMP-9, proinflammatory chemokines (CXCL10 and CCL2), and cytokines (IL-1β) after IR injury. In parallel with the role of sevoflurane preconditioning in the course of IR, we found similar profiles in rats intrathecally pretreated with siRNA, suggesting that sevoflurane regulates MMP-9-evoked pronociceptive actions after IR. These results confirm and extend previous studies of preconditioning induced by volatile anesthetic agents [17]-[20].

Microglial cells are key responders to neural infarcts and play a central role in regulating the pathogenesis of IR injury [3],[4],[10],[28]. As the resident macrophages of the spinal cord, they retract and thicken their cellular processes to transform into an amoeboid form within 30 min of neural insult and migrate toward vulnerable areas to confine the lesion and are capable of phagocytosis [3],[7],[11],[28]. However, some studies suggest that microglia activation has both positive and negative effects in various neurological models and conditions [28]. In addition to releasing neurotrophic factors, activated microglial cells also secrete various pro-inflammatory cytokines, nitric oxide, and other factors that can aggravate injury [3],[11],[28],[29]. How the specific timing and extent of microglial activation might affect these deleterious versus beneficial consequences for spinal cord recovery remains a controversial topic. Karwacki Z and colleagues reported that sevoflurane in concentrations of 1 and 2 MAC (minimum alveolar concentration) inhibited ischemia-induced polymorphonuclear neutrophil (PMN) adhesion. With similar immunophenotypical features as those of PMN, sevoflurane was demonstrated to modulate microglial adhesion [30]. Herein, we used Iba-1 as cellular marker of microglia and analyzed microglial activation qualitatively and quantitatively. Our data show that after IR injury, microglia were rapidly activated in IR group rats, as cells were intensely stained with Iba-1 antibody and possessed enlarged and vacuolated cell bodies extending stout processes (Figure4A). Sevoflurane preconditioning markedly prevented increases in Iba-1 protein expression (Figure5A), completely consistent with the results of Karwacki Z and colleagues [30].

In case of IR, infiltration and migration of microglia into the spinal cord have been linked with injury to the neurovasculature caused by substances released from activated microglia, namely matrix metalloproteinases (MMPs) [7],[14]. MMPs are capable of degrading extracellular matrix components and are responsible for promotion of BSCB opening [9],[14],[30]. Among MMPs, MMP-9 has been reported to be the most abundant inducible factor in the pathology of ischemia in several previous studies [8],[9],[11],[31] including our rabbit IR model [6]. In the present study, we also found that fluorescently labeled MMP-9 colocalized with Iba-1-positive cells in the injured spinal cords of rats subjected to IR injury, and sevoflurane preconditioning prevented the increased number of such double-labeled microglia (Figure4A). Correspondingly, western blot analysis (Figure5A) confirmed the results of the immunofluorescence analysis. Representative immunoblots showed that reduced MMP-9 protein expression was associated with low expression of Iba-1, 36 h after surgery. These results were consistent with the study of Matsmumoto and colleagues [32], who also found that the initial changes in MMP activity could be detected as early as 15 min after insult and persisted for up to 48 h; a time frame that exactly coincides with the morphological and functional changes of microglia [12],[32]. As expected, in the present study, intrathecal infusion of MMP-9 siRNA had a suppressive effect on the number of double-labeled activated microglia and on MMP-9 activity. In addition to extracellular remodeling of the BSCB, MMP-9 has anti-apoptotic and anti-inflammatory effects in various models [8],[9],[32]-[34]. It is well known that apoptosis is one of the major mechanisms that leads to cell death after IR. After ischemia, DNA fragmentation and chromatin condensation function as pro-apoptotic signals that trigger caspase activation [22],[25],[26]. The present study also demonstrated that sevoflurane preconditioning influences motor function by likely inhibiting infiltration and migration of microglia to injured neurons, in accordance with the phenomena described by Choi MS and colleagues that microglia, the resident effectors of the CNS, migrate into injured sites to initiate neuronal inflammation after insults [23]. Moreover, we performed TUNEL staining and measured protein levels of cleaved caspase-3, a known marker of apoptosis within neurons, to investigate the mechanisms by which sevoflurane might attenuate the development of neuronal apoptosis after IR (Figure2A, B). The data showed that TUNEL-positive cells mainly displayed a neuronal morphology and were distributed in the lamina of the anterior horn with increased cleaved caspase-3 immunoreactivity, which were further confirmed by similar changes in caspase-3, Iba-1, and MMP-9 protein levels observed in western blots. Furthermore, downregulation of MMP-9 expression by intrathecal pretreatment with siRNA had similar anti-apoptotic effects as those of sevoflurane preconditioning, suggesting that microglia produce the majority of MMP-9 in response to IR injury, ultimately leading to neuronal apoptosis and motor function impairments.

In regard to inflammation, MMP-9 facilitates migration of inflammatory cells by cleaving inflammatory cytokines and chemokines in several inflammatory diseases, such as Alzheimer's disease, Parkinson disease, and multiple sclerosis, or after stroke or spinal cord trauma [23]. CXCL10 and CCL2 are involved in blood cell and PMN chemotaxis [13],[35] [37]. Moreover, a few studies also have reported that CXCL10 and CCL2 were chemokines that acted as potent recruiters with their primary effects on the innate immune system via chemotaxis residual microglia during peripheral and CNS inflammation [35],[36],[38]. Our study suggests that IR-induced upregulation of MMP-9 expression drives production of proinflammatory chemokines (CXCL10 and CCL2) and cytokines (IL-1β) that in turn amplify inflammatory responses, which ultimately promote neuronal apoptosis leading to hind-limb motor dysfunction [23],[37]. Thus, decreased expression could contribute to reduced infiltration or migration of resident microglia of the spinal cord, leukocytes, and other proinflammatory cells in circulation through the disrupted BSCB, thereby ameliorating the inflammatory response [35]-[37],[39]. Notably, prophylactic administration of neutralizing antibodies against CXCL10 and CCL2 prevents microglial infiltration, decreases neuronal loss, and improves behavioral recovery from spinal cord injury [37]. Further, in support of the above hypothesis, intrathecal pretreatment with specific siRNAs targeting MMP-9 had similar effects on reducing the release of chemokines (CXCL10 and CCL2) and products of proinflammatory cytokines ( IL-1β) leading to decreased neuronal loss and improved behavioral recovery from spinal cord injury (Figures 2 and 4). This inhibition by MMP-9 siRNA may limit the extent of BSCB disruption as well as the extent of microglial activation and progressive recruitment to the spinal cord, which correlated with caspase 3 expression and inflammation-mediated neuronal loss [15]. One the basis of published work, IL-1β is well-recognized as proinflammatory proteins that are associated with activated microglia [3],[4],[35]. Meanwhile, IL-1β is activated via cleavage from its precursor by caspase-1 (the IL-1β converting enzyme), as well as by other enzymes such as MMP-9 and MMP-2 [15]. Consistent with previous studies, the data in Figure5 also show that IL-1β expression was increased in the spinal cord in a microglial activity-dependent manner during the course of IR [4],[10]. Intrathecal MMP-9 siRNA treatment abrogated the increase in IR-induced IL-1β content, suggesting that IL-1β activation was a downstream mechanism underlying MMP-9-induced neuroinflammation accompanied by microglial activation [13]. Thus, it is conceivable that MMP-9 might antagonize these inflammatory factors via inhibiting the release of chemokines (CXCL10 and CCL2) and proinflammatory cytokines (IL-1β), which is in agreement with recent studies by Kjell J and Huang CY [38],[40]. They further suggested that neurons and microglia form a highly interdependent network that is now considered to be the functional unit of the CNS. Although it is unknown whether microglia are the first responders to injury or respond to an "emergency signal" from neurons, it is conceivable that neuronal apoptosis following IR injury combined with extracellular microenvironmental imbalances might be the major stimuli for microglial activation [4],[28],[31]. All the above evidence, when considered together, suggests that upon IR-induced activation, microglia express increased CXCL10, CCL2, and other chemotaxis-associated molecules, which leads to the recruitment of inflammatory cytokines to the lesional area, resulting in amplified activation of additional spinal cord-derived microglia leading to neuronal loss. Blocking this pathological sequence by sevoflurane preconditioning had beneficial effects on BSCB integrity and hind-limb functional recovery. Nevertheless, it's worth noting that that the neuroprotective effects of sevoflurane preconditioning against spinal ischemia is very complicated. Some other underlying mechanisms such as activation of extracellular signal-regulated kinase or potassium ATP channel, as well as through free radicals-mediated up-regulation of antioxidant enzymes were also demonstrated to play a role [16]-[18],[26],[41]. Further studies focused on the links between different actions of sevoflurane preconditioning still need to be conducted.

In conclusion, the current study provides the first direct evidence that sevoflurane preconditioning produces anti-inflammatory and anti-apoptotic effects on spinal cord IR injury in rats. We provide additional evidence that this effect is possibly mediated through the inactivation of microglia resulting in reduced expression of MMP-9 and proinflammatory chemokines and cytokines. These data suggest that sevoflurane preconditioning may provide a new practical method for protecting against perioperative spinal cord IR injury.

X-QL and X-ZC participated in the animals' care and prepared all the animal models. X-QL, X-ZC, JW, and W-FT participated in tissue preparation and sectioning, and performed most of the immunohistochemistry; X-QL, BF, X-ZC, and W-FT performed western blotting assays and statistical analysis; HM and JW were involved in guiding model design and study design; JW and HM gave important direction and input for data analysis and manuscript writing. All authors read and approved the final manuscript.