Suppressing HMGB1-TLR4-Mediated Neuroin ammation Alleviates Morphine Tolerance via Inhibiting AMPK-HO-1 Pathway in The Spinal Cord of Mice

Tong-Tong Lin Nanjing Medical University Lei Sheng Nanjing Medical University Li Wan Nanjing Medical University Lu-Lu Ji Nanjing Medical University Jin-Can Li Nanjing Medical University Xiao-Di Sun The First A liated Hospital of Nanjing Medical University Yin-Bing Pan The First A liated Hospital of Nanjing Medical University Chen-Jie Xu Nanjing First Hospital Liang Hu Nanjing Medical University Xue-Feng Wu Nanjing University Yuan Han Eye & ENT Hospital Fudan University Chun-Yi Jiang (  jcy@njmu.edu.cn ) Nanjing Medical University Wen-Tao Liu Nanjing Medical University

immune response [6]. It was reported that morphine could trigger TLR4-mediated neuroin ammation by binding with MD-2, a TLR4 accessory protein [7]. Blockade of TLR4 inhibited the activation of microglia and attenuated morphine tolerance, which suggested that TLR4 is very important in the development of morphine tolerance [8].
There are several endogenous TLR4 agonists, such as heat shock proteins (HSPs), high-mobility group box 1 (HMGB1), heparan sulfate (HS) [9] and brinogen (Fg) [10]. Our previous study indicated that morphine could induce the release of HSP70 from neurons. As an important DAMPs, HSP70 induced the activation of TLR4-NLRP3 in ammasome signal pathway and led to neuroin ammation [11]. Furthermore, it was reported that disul de high mobility group box-1 (ds-HMGB1) could mimic the effects of morphine induced-persistent sensitization through TLR4 receptor (Grace et al., 2016). The released extracellular HMGB1 could increase the expression of proin ammatory cytokines, such as IL-1β, IL-6 and TNF-α [12]. HMGB1, a non-histone DNA binding protein, mainly located in the cell nucleus could stabilize nucleosome formation, and regulate the interaction of transcription factors with DNA in the physiological state [13]. It was reported that signi cant amount of HMGB1 was released from the nucleus into the extracellular space after neuronal injury immediately [14]. HMGB1, as an important DAMPs has been shown to participate in neuroin ammation, modulate autophagy and apoptosis as well as regulate gene transcription [15,16]. It is considered as a critical factor in several pathophysiological conditions. Active HMGB1 secretion by in ammatory cells and passive release by dying neurons affects cell viability and apoptosis, and contributes to neuroin ammation and neurodegeneration associated with Parkinson disease [13]. The release of HMGB1 from astrocytes through a pannexin-1 and P2 × 7 receptor signaling cascade could then contribute to the development of major depressive disorder (MDD) [17].
These information give us a hint that HMGB1 might be released by chronic morphine administration and the extracellular HMGB1 could be another critical morphine-induced in ammatory mediator contributing to the development of morphine tolerance. Here, we investigate the role of HMGB1 in pathogenesis of morphine tolerance as well as its underlying mechanisms.

Animals
Adult male ICR mice (18-22 g) and adult male Sprague-Dawley rats (200-250 g) were provided by the Experimental Animal Center at Nanjing Medical University, Nanjing, China. Animals were housed 5 to 6 per cage under pathogen-free conditions with soft bedding under controlled temperature (22±2˚C) and a 12-h light/dark cycle (lights on at 8:00 am). Behavioral testing was performed during the light cycle (between 9:00 am and 5:00 pm). The animals were allowed to acclimate to these conditions for at least 2 days before starting experiments. For each group of experiments, the animals were matched by age and body weight.

Chemicals and Reagents
Morphine hydrochloride was purchased from Shenyang First Pharmaceutical Factory, Northeast Pharmaceutical Group Company (Shenyang, China). Glycyrrhizin was purchased from Sigma Aldrich (St. Louis, MO, USA). Compound C and TAK242 were purchased from MedChemExpress (New Jersey, USA).

Tolerance model and behavioral analysis
We habituated animals in the testing environments for 2 days and carried out behavioral testing in a blinded manner. For the test of chronic tolerance, mice were intrathecally injected with vehicle or morphine (10 μg/10 μL) once daily for seven consecutive days. Behavioral testing was performed 30 min after morphine administration by tail-ick assay every morning. The tail-ick test was performed using a water bath with temperature maintained at 52°C. Each animal was gently wrapped in a cloth by the experiment. The distal one-third of tail was immersed in a water bath, and mice rapidly removed their tail from the bath at the rst sign of the discomfort. Stop the chronometer as soon as the mouse withdraws its tail from the hot water and record the latency time (in sec). A cut-off time of 10 s was set to avoid tissue damage. Different doses of Glycyrrhizin (25, 50, 100 mg/kg) were administered by intragastric injection 15 min before morphine administration. Different doses of Compound C (1, 3 or 10 μg/10 μL) were administered by intrathecal injection 15 min before morphine administration. Data were calculated as a percentage of maximal possible effect (%MPE), which was calculated by the following formula: 100% × [(Drug response time -Basal response time) / (10 s -Basal response time)] = %MPE. Rats were intrathecally injected with morphine (20 μg/10 μL) daily for 7 consecutive days.
BV-2 cells were maintained in humidi ed 5% CO2 at 37 °C in Dulbecco's Modi ed Eagle's Medium (DMEM, KenGEN BioTECH, China) supplemented with 10% (v/v) FBS (Gibco), 80 U/mL penicillin, and 0.08 mg/mL streptomycin. For further experiments, BV-2 cells were plated in 6-well plate overnight and then treated with recombinant HMGB1 protein or conditional medium from SH-SY5Y cells. Then the cell extracts and precipitated supernatants were analyzed by immunoblot assay.

Intrathecal injection procedure
To perform intrathecal (i.t.) injections, the mice were placed in a prone position and the midpoint between the tips of the iliac crest was located. A Hamilton syringe with a 30-gauge needle was inserted into the subarachnoid space of the spinal cord between the L5 and L6 spinous processes. Proper intrathecal injection was systemically con rmed by observation of a tail ick. Intrathecal injection did not affect baseline responses, compared with latencies recorded before injection.

Immunohistochemistry
For uorescence immunohistochemistry, mice were anaesthetized and transcardially perfused with 4% cold paraformaldehyde on day 7. Lumbar spinal cords were harvested, post-xed for 4 h at 4°C in 4% paraformaldehyde, and then cryo-protected sequentially in 10, 20 and 30% sucrose overnight for 3 days.  (1:200) were incubated with the tissue section in 10% normal donkey serum and 0.01% Triton-X-100 overnight at 4°C. For c-fos and CGRP uorescence immunohistochemistry, primary antibodies against c-fos (1:300) and CRGP (1:300) were incubated with the tissue section in 10% normal donkey serum and 0.01% Triton-X-100 overnight at 4°C. The appropriate uorescent secondary antibody (1:300, Alexa Fluor 488 or 567) was used for each primary antibody. Confocal microscopy of immuno uorescence in the dorsal horn was performed with a confocal microscopy (Zeiss LSM710, Germany).

RNA interference
HO-1 siRNA (EHU051241), AMPK siRNA (AMPKα1, EHU074041; AMPKα2, EHU042081) and control siRNA (EHUEGFP) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Control siRNA was used as a negative control. For the transfection of siRNA, SH-SY5Y cells were cultured in 6-well plates with antibiotic-free medium the day before transfection. The transfection was conducted when cells reached 50 ~ 70% con uence using Lipofectamine 2000 (Invitrogen, USA) and serum-free medium according to the manufacturer's instructions. After 5 h, the transfection medium was replaced with the culture medium containing 10% FBS and then incubated at 37 °C.

Collection of cerebrospinal uid (CSF)
Adult male Sprague-Dawley rats (200-250 g) were housed under a 12 light/dark cycle, with food and water available ad lib. The animals were anesthetized with pentobarbital sodium (50 mg/kg, i.p.). The CSF was carefully collected from the cisterna magna of each rat and inspected for blood contamination. Contaminated samples were discarded. Approximately 80 μL of CSF was collected from each animal. After a short centrifugation step (5 min at 5000 g, 4 °C), the samples were dissolved in 2×SDS loading buffer, boiled, and analyzed by SDS-PAGE followed by western blotting.

Statistical Analysis
GraphPad Prism 7 software (GraphPad Software, San Diego, Calif.) was used to conduct all the statistical analyses. The differences between two groups were evaluated by one-sample t-test. The data from more than two groups were evaluated by one-way ANOVA or two-way ANOVA. Results were represented as mean ± SEM of the independent experiments. Results described as signi cant were based on a criterion of P < 0.05.

Morphine induces the release of HMGB1 in vivo and in vitro
We established morphine tolerance model with SD rats collecting the CSF to investigate whether morphine could induce the e ux of HMGB1 into extracellular environment. As shown in Figure 1A, the immunoblot data indicated that morphine markedly caused HMGB1 to be released into CSF. To further investigate the cellular mechanism underlying morphine-induced release of HMGB1, analysis of the cellular distribution of HMGB1 by confocal microscopic scanning was performed. It was showed that, in naive mice, HMGB1 mainly co-localized with NeuN (neuronal marker), but not with Iba1 (microglia marker) or GFAP (astrocyte marker) in the spinal cord (Fig. 1B). Therefore, we utilized human neuroblastoma cell line SH-SY5Y cells, which express both mu-and delta-opioid receptors (ratio approximately 4.5:1) [18], to further con rm the release of HMGB1 induced by morphine in neuron. We incubated SH-SY5Y cells with different concentrations (50, 100, 200 μM) of morphine for 12 h (the cells were stimulated as described) [2], then the supernatants of SH-SY5Y cells were collected and analyzed by western blot. MTT assay indicated that the different concentrations (50, 100, 200 μM) of morphine did not affect cell proliferation (Additional le 1: Figure S1). We found that morphine promoted the e ux of HMGB1 into extracellular environment in a concentration-dependent manner (Fig. 1C). Furthermore, the immuno uorescence results showed that morphine caused the migration of HMGB1 rstly from nucleus to cytoplasm before releasing to extracellular environment (Fig. 1D).
Glycyrrhizin attenuates chronic morphine tolerance According to the data mentioned above, morphine induced the release of HMGB1. Then we questioned whether the released HMGB1 was important for the development of morphine tolerance. As a selective HMGB1 inhibitor, glycyrrhizin was utilized to investigate the therapeutic effects to the released HMGB1 in morphine tolerance. Glycyrrhizin did not affect the acute analgesic effect of morphine ( Fig. 2A), and the behavioral test results showed that glycyrrhizin attenuated chronic morphine tolerance in dose-dependent manner (Fig. 2B). At day 7, the MPE at 30 min after morphine administration decreased to 14.5% in chronic morphine-treated mice, whereas mice with co-administration of glycyrrhizin (25, 50, and 100 mg/kg) and morphine displayed the MPE of 26.3%, 35.9% and 54.7%, respectively (Fig. 2B). Furthermore, the expression of c-fos and CGRP protein, two of the immediate early genes rapidly expressed in neurons after a noxious stimulus, were increased in the dorsal horn of the spinal cord after morphine tolerance, and these increases were then suppressed by glycyrrhizin (Fig. 2C). Extracellular HMGB1 triggers in ammatory response dependent on TLR4 in microglia.
Next, we investigated the role of extracellular HMGB1 in neuroin ammation. Microglia, accounting for 10%~15% of all central nerve cells, are natural immune cells in the central nervous system. When exposed to a variety of damaging stimuli, such as ischemia/hypoxia, trauma and infection, microglia cells can be activated rapidly and release a large number of proin ammatory cytokines, including TNF-α IL-1β and IL-6 [19][20][21]. TLR4 can be activated by DAMPs, triggering its downstream signaling pathway, consequently evoking innate immune response through the maturation of IL-1β. As HMGB1 is an important TLR4 agonist, we investigated the effect of extracellular HMGB1 on in ammatory response and immortalized murine microglial cell line BV-2 was utilized [22,23]. Firstly, we found that repeated morphine administration led to the phosphorylation of NF-κB p65 and the upregulation of proin ammatory cytokines IL-1β in the spinal cord ( Fig. 3A and B). Glycyrrhizin signi cantly attenuated the phosphorylation of NF-κB p65 and the expression of IL-1β in the spinal cord ( Fig. 3A and B). Furthermore, Immuno uorescence data showed that repeated morphine treatment (10 μg/10 μL, once daily for 7 days) led to the activation of microglia (Iba-1 as microglia marker), and glycyrrhizin signi cantly inhibited morphine-induced activation of microglia (Fig. 3C). In order to verify whether the extracellular HMGB1-triggered in ammatory response is dependent on TLR4 in microglia, we found that recombinant HMGB1 (25 nM) signi cantly upregulated the phosphorylation of NF-κB p65 in BV-2 cells. In addition, the administration of TLR4 antagonist (TAK242, 10 μM) suppressed the upregulation of the phosphorylation of NF-κB p65 caused by recombinant HMGB1 in BV-2 cells (Fig. 3D). The data indicated that TLR4 was essential for the in ammatory response caused by HMGB1. In order to further con rm the role of HMGB1 in inducing in ammatory response, we utilized conditional medium (CM) from morphinetreated (200 μM, 12 h) SH-SY5Y cells to activate BV-2 cells. Anti-HMGB1 antibody (2 μg/μL) suppressed CM-induced upregulation of phosphorylation of NF-κB p65 and the expression of IL-1β ( Fig. 3E and F). Normal IgG (2 μg/μL) did not show an inhibitory effect (Fig. 3E and F).
The release of HMGB1 induced by morphine depends on AMPK/HO-1 pathway The next question addressed how morphine induced the release of HMGB1 from neurons. HO-1, a stressresponsive protein, serves a vital metabolic function as the rate-limiting step in the degradation of heme [24]. It may function as a pleiotropic regulator during in ammation. Increased HO-1 expression in tissue is commonly associated with increased in ammation or oxidative stress. HO-1 can be up-regulated by a variety of factors, including hypoxia, hyperoxia, heat shock, cytokines, heavy metals, hydrogen peroxide, UV irradiation, and its substrate, heme [25,26]. It was reported that HO-1 signal could prevent HMGB1mediated activation of NLRP3 in ammasomes. HMGB1-mediated in ammatory and immune reactions in nervous system played an essential role in the generation of neuroin ammation. [27,28]. However, despite these observations, deleterious consequences of HO-1 overexpression have been reported in vitro and in vivo. Song W et al. reported that in rat astrocytes, HO-1 over-expression resulted in signi cant oxidative damage to mitochondrial lipids, proteins, and nucleic acids, partial growth arrest, and increased cell death [29]. Besides, glial HO-1 over-expression was considered as a possible cause of pathological iron deposition in the aging and degenerating mammalian CNS [30]. Furthermore, HO-1 activity inhibitors displayed the therapeutic value in Alzheimer's disease and related neurodegenerative disorders [31].
Recent studies indicated that HO-1 can be regulated under several mechanisms, such as, SIRT-1, AMPK and Nrf-2 [32,33]. Among them, AMPK, a key regulator of energy homeostasis, plays a fundamental role in the process of chronic pain occurrence, development and maintenance [34]. AMPK has recently emerged as a novel target for the treatment of pain with the exciting potential for disease modi cation [35]. Zhang et al. demonstrated that morphine activated AMPK pathway, induced epithelial-mesenchymal transition (EMT) via upregulating Snail and Slug levels, and increased oxidative stress in esophageal carcinoma cells [36]. In addition, morphine could also inhibit PINK1/Parkin-mediated mitophagy in spinal cord neurons and increase the phosphorylation of AMPK [37]. Hence, we speculated that HO-1-HMGB1 activation was downstream to AMPK. Firstly, we found morphine induced the phosphorylation of AMPK in a concentration-dependent manner in SH-SY5Y cells and Compound C signi cantly decreased the level of HO-1and released HMGB1 induced by morphine in SH-SY5Y cells (Fig. 5A-C). Then we utilized AMPK siRNA, it downregulated the level of phosphorylation of AMPK, HO-1 and released HMGB-1 of SH-SY5Y cells (Fig. 5D-F).
Furthermore, to further con rm our pharmacological results, repeated i.t. injection of Compound C was perform to evaluate the effect of AMPK in chronic morphine tolerance in vivo. Compound C coadministration with morphine signi cantly potentiated acute morphine analgesic effect and attenuated chronic morphine tolerance in mice ( Fig. 6A and B). At day 7, the MPE at 30min after morphine administration decreased to 14.39% in chronic morphine-treated mice, whereas mice co-administration with Compound C (1, 3, 10 μg/10 μL) and morphine displayed the MPE of 28.79%, 51.52% and 70.05% respectively (Fig. 6B). Furthermore, western blot data showed that Compound C could decreased the release of HMGB1 in CSF (Fig. 6C) and suppressed the level of p-AMPK HO-1. Furthermore, Compound C inhibited the phosphorylation of p65 induced by morphine in the spinal cord ( Fig. 6D-F). These data provided su cient evidence that morphine induced the release of HMGB1 via AMPK/HO-1 pathway and the suppression of this pathway could effectively improve chronic analgesic tolerance induced by morphine. Figure 6 Discussion In the present study, the principal ndings are: (1) administration of morphine induces the release of HMGB1, an endogenous TLR4 ligand, from neurons; (2) morphine induces the release of HMGB1 to extracellular environments via AMPK/HO-1 pathway; (3) extracellular HMGB1 activates microglia and causes in ammatory response, upregulating the phosphorylation of p65 and the levels of IL-1β.
The mechanisms underlying morphine tolerance is complex and far beyond having a clear explanation. Several mechanisms are involved in the tolerance to analgesic opioids, including desensitization or internalization of the opioid receptor, elevation of cAMP levels and downregulation of spinal glutamate transporters [38]. Especially, compelling evidence has demonstrated that chronic morphine exposure induced sterile neuroin ammation in the spinal cord. HMGB1 has been implicated as a key factor in the mediation of neuroin ammation processes in several pathophysiological conditions. For example, HMGB1 is massively released into the extracellular space immediately after ischemic insult and it subsequently induces neuroin ammation in the postischemic brain [39]. HMGB1-Mac1-NADPH oxidase signaling axis bridges chronic neuroin ammation and progressive dopaminergic neurodegeneration in PD progression [40]. In addition, the persistent release of HMGB1 contributes to tactile hyperalgesia in a rodent model of neuropathic pain [41]. Anti-HMGB1 monoclonal antibody (mAb) has been shown to be effective for the treatment of a wide range of CNS diseases when modeled in animals, including stroke, traumatic brain injury, epilepsy and Alzheimer's disease [42]. Recently, Qian J et al. discovered that morphine-mediated upregulation of HMGB1 in the spinal cord which contributed to analgesic tolerance and hyperalgesia [43]. In our study we not only con rmed that morphine could induce the e ux of HMGB1 into extracellular environment via AMPK-HO-1 axis. But also, we illustrated that the released HMGB1 triggered TLR4 signaling which consequently induced the upregulation of the phosphorylation of NF-κB p65 and proin ammatory cytokines. Our immunoblot data gave the evidence that morphine markedly caused HMGB1 released into CSF under the tolerance model with SD rats (Fig. 1A). Furthermore, the immuno uorescence results and immunoblot data showed that morphine induced the release of HMGB1 from SH-SY5Y cells in a concentration-dependent manner (Fig. 1C and 1D). Therefore we provided a hypothesis that the released HMGB1 induced by morphine probably played an important role in neuron-microglia crosstalk, especially in the development of neuroin ammation. Previous studies revealed that HMGB1 presented its activities by serving as a ligand for several different receptors, including TLR2, TLR4, TLR5, TLR9, and RAGE [12]. Among them, HMGB1-triggered TLR4 signaling is a strong inducer of proin ammatory cytokine production. In our study, we demonstrated that extracellular HMGB1 upregulated the phosphorylation of NF-κB p65 and increased the transcription of IL-1β in the spinal cord ( Fig. 3A and 3B). Recombinant HMGB1 protein signi cantly upregulated the phosphorylation of NF-κB p65 in BV-2 cells and the administration of TLR4 antagonist suppressed the upregulation of the phosphorylation of NF-κB p65 in BV-2 cells (Fig. 3D). In addition, anti-HMGB1 antibody could signi cantly inhibit the upregulation of the phosphorylation of NF-κB p65 and IL-1β in BV-2 cells in protein level (Fig. 3E). Our in vitro study showed a direct effect of morphine in inducing HMGB1 release from neurons and the released HMGB1 displayed its ability in triggering neuroin ammation mediated by TLR4. Furthermore, glycyrrhizin (HMGB1 inhibitor, 100 mg/kg) could increase MPE by 39.88% comparing with that in the morphine-treated group at day 7 (Fig. 2B). Immuno uorescence staining data revealed glycyrrhizin could decreased the level of CGRP (Fig. 2C), which was a peptide released by primary afferents and was able to mediate the activation of NMDA receptor in neurons. And glycyrrhizin also downregulated c-fos (Fig. 2C), which were implicated in pain transmission and morphine tolerance.
Next, we explored the mechanism of HMGB1 releasing from neurons caused by morphine. It is increasingly recognized that HO-1 down-regulates the in ammatory response. In a rat model of pleurisy, up-regulation of HO-1 signi cantly attenuates acute cellular in ammation [44]. In addition, numerous studies indicated that HO-1 decreased the expression of HMGB1 during in ammatory response. Although many preclinical research studies have pointed to anti-in ammatory effects of HO-1 in tissue injury, recent studies also propose a pro-pathogenic effect of HO-1 in the propagation of chronic in ammation. HO-1 has been considered to be among the strongest positive predictors of metabolic disease in humans [45]. For example, although previous studies have shown that systemic induction of HO-1 by chemical inducers reduces adiposity and improves insulin sensitivity. Huang JY et al. reported that HO-1 overexpression in adipocytes does not protect against high fat diet induced obesity and the development of insulin resistance in mice [46]. In addition, hepatocyte and macrophage conditional HO-1 deletion in mice conferred protection against diet-induced insulin resistance and in ammation, dramatically reducing secondary disease such as steatosis and liver toxicity [47]. Heterozygous Hmox1 knockout mice were also shown to be protected against high fat diet induced insulin resistance by reducing macrophage migration [48].
Therefore, the role of HO-1-HMGB1 axis in the development of morphine tolerance remains complex and incompletely understood. Our results indicated that morphine induced the release of HMGB1 via the activation of HO-1 and the utilization of Znpp and HO-1 siRNA signi cantly decreased the release of HMGB1 (Fig. 4B and D) induced by morphine. Targeting HO-1 remains an attractive therapeutic strategy for the treatment of in ammatory conditions. In the unfolded protein response, activated AMPK could boost the Nrf2/HO-1 signaling axis [49]. Esculentoside A has protective potential against APAP toxicity in acute liver failure by potentiating the Nrf2/HO-1-regulated survival mechanism through the AMPK/Akt/GSK3β pathway [50]. Here, we focus on AMPK based on its important role in the regulation of pain, including neuropathic pain, incision-induced acute and chronic pain [51][52][53]. According to the mentioned above, we inferred that the release of HMGB1 induced by morphine probably depends on AMPK-HO-1 axis. Our data demonstrated that Compound C and AMPK siRNA could suppress the elevation of HO-1 and the releasing of HMGB1 induced by morphine in SH-SY5Y cells (Fig. 5). In vivo experiments, Compound C displayed the ability to effectively improve morphine tolerance through the suppression of HMGB1 releasing to CSF.

Conclusion
Our study indicated that morphine-induced extracellular HMGB1 was critical for morphine tolerance. We demonstrated that morphine could induce the release of HMGB1 from neurons via AMPK-HO-1 pathway.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.