Disulfide-Containing High Mobility Group Box-1 Promotes N -Methyl- d -Aspartate Receptor Function and Excitotoxicity by Activating Toll-Like Receptor 4-Dependent Signaling in Hippocampal Neurons

Abstract

Aims: Using primary cultures of mouse hippocampal neurons, we studied the molecular and functional interactions between high mobility group box-1 (HMGB1) and the N-methyl-d-aspartate receptor (NMDAR), two proteins playing a key role in neuronal hyperexcitability. By measuring NMDA-induced calcium (Ca²⁺) increase in neuronal somata and neurotoxicity as functional read-out parameters, we explored the role of the redox state of HMGB1, the receptor involved, and the molecular signaling underlying its interactions with postsynaptic NMDAR. We also investigated whether HMGB1 redox state affects its proconvulsive effects in mice. Results: Nonoxidizable HMGB1 with a triple cysteine-to-serine replacement (3S-HMGB1) was ineffective on NMDA response. Conversely, the disulfide-containing form of HMGB1 dose dependently enhanced NMDA-induced Ca²⁺ increase in neuronal cell bodies. This effect was prevented by BoxA, a competitive HMGB1 antagonist, and by Rhodobacter sphaeroides lipopolysaccharide (LPS-RS), a toll-like receptor 4 (TLR4) selective antagonist, and it was abrogated in neurons lacking TLR4 while persisting in the absence of receptor for advanced glycation end products (RAGE). TLR4 and NMDAR subunit 1 (NR1) and 2B (NR2B) were colocalized in neurons. Disulfide HMGB1 effect on NMDA-induced Ca²⁺ influx was prevented by 3-O-methylsphingomyelin (3-O-MS) and 4-amino-5-(4-chlorophenyl)-7-(t-butyl) pyrazolo [3,4-d] pyrimidine, (PP2) selective inhibitors of neutral sphingomyelinase and Src-family Tyr kinases, respectively. Disulfide HMGB1, but not 3S-HMGB1, increased Tyr¹⁴⁷² phosphorylation of the NR2B subunit of the NMDAR, which is known to increase Ca²⁺ channel permeability. Similarly, disulfide HMGB1 increased NMDA-induced neuronal cell death in vitro and enhanced kainate-induced seizures in vivo. Innovation and Conclusion: We describe a novel molecular neuronal pathway activated by HMGB1 that could be targeted in vivo to prevent neurodegeneration and seizures mediated by excessive NMDARs stimulation. Antioxid. Redox Signal. 21, 1726–1740.

Introduction

High mobility group box-1 (HMGB1) is a ubiquitous chromatin-binding protein. In addition to this nuclear location, HMGB1 can rapidly translocate into the cytoplasm after infectious or sterile injuries; then, it is released extracellularly. Extracellular HMGB1 acts as a danger signal (i.e., alarmin) by alerting cells in the microenvironment of an imminent or ongoing tissue threat/damage, thus activating homeostatic programs (5). In this context, HMGB1 activates vascular endothelial cells, recruits leukocytes to the site of injury, and promotes the production of cytokines and reactive oxygen species (ROS) (2, 5, 6, 47, 56). These activities are mediated by CXCR4, the receptor for advanced glycation end products (RAGE) and toll-like receptors (TLR) (45). Recent evidence demonstrates that the redox state of HMGB1 is pivotal for determining which receptor is activated and its functional effects (55, 61, 69).

Innovation

High mobility group box-1 (HMGB1) in its oxidized (disulfide) form, most likely determined by the redox state of the extracellular milieu, potentiates N-methyl-d-aspartate (NMDA)-mediated calcium (Ca²⁺) influx in pyramidal neuron cell bodies. HMGB1 effect is mediated by activation of neuronal toll-like receptor 4 (TLR4) colocalized with NMDA receptors (NMDARs). We demonstrate that HMGB1-TLR4 axis triggers neutral sphingomyelinase and Src kinases activities, and the phosphorylation of the NMDAR subunit 2B. This rapid onset of post-translational pathway underlies HMGB1–NMDAR interaction, leading to intracellular Ca²⁺ increase and contributing to seizures and cell loss. The HMGB1-TLR4 pathway, therefore, represents a new target to prevent central nervous system (CNS) dysfunctions that are mediated by excessive NMDAR stimulation.

HMGB1 is a redox-sensitive protein containing three cysteines, namely C23, C45, and C106. C23 and C45, which are in the first HMG-box domain of HMGB1 (BoxA), can readily form a disulfide bond (48). C106, within the second HMG-box domain of HMGB1 (BoxB), is unpaired and required for HMGB1 binding to toll-like receptor 4 (TLR4) (68). Intracellular HMGB1, either nuclear or cytosolic, is in a reduced state (22), while on its extracellular release, it can form the disulfide bond (62, 69). HMGB1 containing the C23–C45 disulfide bond and free C106 have proinflammatory activity that is mediated by TLR4, thus inducing cytokine production and release from immunocompetent cells (68, 69). In contrast, HMGB1 containing all reduced cysteine residues (all-thiol HMGB1) has chemotactic properties that are mediated by the formation of a heterocomplex with the chemokine CXCL12, which activates the CXCR4 receptor (61, 62). Both HMGB1 activities can be irreversibly inactivated by sulfonation of cysteine residues, which probably represents a physiological mechanism to terminate the signaling of extracellular HMGB1 (69).

In addition to its immune-related functions, HMGB1 acts on brain cells by promoting neuronal differentiation, neurite outgrowth, and glia activation (29, 42 –44). Recently, we described a key role of HMGB1 in the generation and recurrence of seizures in animal models of epilepsy, thus unveiling its neuromodulatory effects in vivo (26, 37). HMGB1-induced neuronal hyperexcitability in seizure models was chiefly mediated by TLR4 and involved N-methyl-d-aspartate receptor (NMDAR) (26, 37). NMDAR calcium (Ca²⁺) permeability is increased by the expression and Tyr phosphorylation of the NMDAR subunit 2B (NR2B) (73). Activation of presynaptic NMDAR contributes to excitatory neurotransmission by promoting Ca²⁺-dependent glutamate release (38, 54). A peptide fragment of HMGB1 that lacks receptor-binding properties was shown to interact directly with presynaptic NR2B-containing NMDAR, thus facilitating NMDA-mediated glutamate release (42). Postsynaptic NMDAR expressed in neuronal somata and dendrites plays a critical role in synaptic plasticity (59), but intracellular Ca²⁺ overload due to excessive activation of NR2B-expressing NMDAR results in hyperexcitability and excitoxicity (27).

In order to elucidate the mechanism by which HMGB1 contributes to NMDAR-mediated hyperexcitability and excitotoxicity (26, 37, 72), we asked whether HMGB1 enhances NMDAR-induced Ca²⁺ increase in neuronal somata. We also investigated the impact of the redox state of HMGB1, the identity of its receptors, and the mode of intracellular signaling in neurons.

Results

NMDAR-mediated neuronal Ca²⁺ influx

Figure 1A depicts the dose-dependent increase in intracellular Ca²⁺ evoked by 2.15 min exposure to NMDA in hippocampal neuronal cultures preloaded with fura-2 acetoxymethyl-ester (fura-2AM). While 1 μM NMDA was ineffective, 10 and 20 μM NMDA induced an average 50% and 80% increase in the peak value of intracellular Ca²⁺ (+39% and +64% for area under the curve [AUC], respectively) (p<0.01 vs. control Krebs buffer by one-way analysis of variance [ANOVA]). The maximal effect was reached at 100–500 μM NMDA (p<0.01 vs. control). NMDA-induced intracellular Ca²⁺ increase reflected ion influx through the Ca²⁺-permeable channel, as it was prevented by 5 min preincubation and co-incubation with 10 μM MK801, an open-channel and use-dependent NMDAR antagonist (67) (Fig. 1B). NMDA half maximal effective concentration was ∼11 and ∼15 μM for peak effect and AUC, respectively. In subsequent experiments, we used 10 μM NMDA, unless otherwise indicated, which corresponds to the concentration previously used to study the functional interactions of NMDA and interleukin-1 beta (IL-1β) or HMGB1 in vitro (25, 31, 64).

FIG. 1.

Intracellular Ca²⁺ increase evoked by NMDA in hippocampal neurons. (A) Depicts the dose–response curve (PEAK and AUC) of NMDA-evoked (1–500 μM) intracellular Ca²⁺ increase (ΔF/F) in primary hippocampal neuronal cultures preloaded with fura-2AM. Data are the mean±SEM (n=7–18 values from three independent experiments) expressed as % of maximal NMDA response. (B) Depicts representative traces of the Ca²⁺ response of hippocampal neurons to 10 μM NMDA alone (red) or along with 10 μM MK801 (black), a selective use-dependent open channel blocker. MK-801 was preincubated for 5 min before, and for 2.15 min along with NMDA. Data are the mean±SEM (n=4 values from two independent experiments). **p<0.01 versus Krebs by one-way ANOVA followed by Tukey's test. AUC, area under the curve; ANOVA, analysis of variance; Ca²⁺, calcium; NMDA, N-methyl-d-aspartate; SEM, standard error of the mean. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Effect of HMGB1 redox state on NMDA-induced Ca²⁺ increase

Preexposure of neurons to disulfide HMGB1 (0.004–400 nM) for 5 min did not affect Ca²⁺ levels per se but potentiated the NMDA-induced Ca²⁺ response dose dependently (Fig. 2A, B). Disulfide HMGB1 at 0.004 nM was ineffective, while a maximal increase in peak value and AUC of ∼40% and 83%, respectively, was reached between 0.4 and 400 nM (p<0.01 by two-way ANOVA). Disulfide HMGB1 (40 nM) was unable to promote Ca²⁺ influx when incubated with an ineffective concentration of NMDA (1 μM, Fig. 2A). When neurons were exposed to all-thiol HMGB1 (reduced form) in the presence of 10 μM NMDA, no effect was observed in the maximal increase in peak value while the AUC value was enhanced (data not shown). This suggested that all-thiol HMGB1 might be ineffective, but would start oxidizing in the incubation medium during the course of the experiment to induce its effect. To elucidate, therefore, the impact of the redox state of HMGB1 on NMDA-induced Ca²⁺ influx, we used a mutant form of HMGB1, 3S-HMGB1 (61), in which cysteines have been replaced by serines and cannot be oxidized. Incubation with 3S-HMGB1 (0.004–40 nM) failed to potentiate the NMDA-induced intracellular Ca²⁺ increase (Fig. 2C).

FIG. 2.

Effect of disulfide and 3S-HMGB1 on NMDA-induced intracellular Ca²⁺ increase in hippocampal neurons. Bargrams represent the neuronal Ca²⁺ response (PEAK and AUC) to 10 μM NMDA alone, or in the presence of increasing concentrations (0.004–400 nM) of disulfide-HMGB1 (A) or 3S-HMGB1 (C). Disulfide- and 3S-HMGB1 per se were ineffective (not shown). Representative traces of neuronal Ca²⁺ response to NMDA±disulfide HMGB1 are reported in (B). Data are the mean±SEM (n=6–19 values from 3–4 independent experiments). Neurons were incubated with HMGB1 for 5 min before, and for 2.15 min with NMDA. **p<0.01 versus Krebs; ^†† p<0.01 versus NMDA by two-way ANOVA followed by Tukey's test. HMGB1, high mobility group box-1. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Receptor-mediated effect of disulfide HMGB1

The effect of 40 nM disulfide HMGB1 on the NMDA-mediated Ca²⁺ response of hippocampal neurons was abolished by 100 nM BoxA, an HMGB1 competitive inhibitor (70) (Fig. 3A; p<0.05 by two-way ANOVA), and by 100 μg/ml Rhodobacter sphaeroides lipopolysaccharide (LPS-RS), a selective TLR4 antagonist (Fig. 3B; p<0.01 by two-way ANOVA). These two antagonists per se neither evoked Ca²⁺ responses (not shown) nor affected Ca²⁺ fluxes induced by NMDA alone (Fig. 3A, B).

FIG. 3.

Effect of BoxA and LPS-RS on disulfide-HMGB1 potentiation of the NMDA response. Bargrams represent the effect of 100 nM BoxA (A) or 100 μg/ml LPS-RS (B) on disulfide-HMGB1 (40 nM) potentiation of 10 μM NMDA-induced Ca²⁺ response (PEAK and AUC). BoxA and LPS-RS, HMGB1 and TLR4 antagonist, respectively, were ineffective per se (not shown). Neurons were incubated with the antagonists for 5 min before adding the other drugs (see legend to Fig. 2 for details). Data are the mean±SEM expressed as a percentage of NMDA response (n=6–20 values from three independent experiments). *p<0.05, **p<0.01 versus NMDA alone; ^† p<0.05, ^†† p<0.01 versus HMGB1+NMDA by two-way ANOVA followed by Tukey's test. LPS-RS, Rhodobacter sphaeroides lipopolysaccharide; TLR4, toll-like receptor 4.

To test the involvement of RAGE and TLR4 in mediating disulfide HMGB1 interaction with NMDAR, we used neuronal cultures from Rage^−/− (Fig. 4A) or Tlr4^−/− mice (Fig. 4B). Ca²⁺ increase induced by 10 μM NMDA was similar in Rage^−/− and wild-type neurons (wt, peak 0.17±0.02; AUC 0.22±0.03; Rage ^−/− peak, 0.15±0.01; AUC, 0.22±0.02). Disulfide HMGB1 enhanced the NMDA-induced Ca²⁺ response similarly in Rage ^−/− and wild-type hippocampal neurons (wt, peak +39%, AUC +82%; Rage ^−/− peak +35%, AUC +77%, Fig. 4A). The potentiation of NMDA response induced by disulfide HMGB1 in Rage ^−/− neurons was abolished by 100 μg/ml LPS-RS (Fig. 4A), as in wild-type neurons (Fig. 3B).

FIG. 4.

Disulfide HMGB1 potentiation of NMDA-induced Ca²⁺ response in hippocampal neurons involves TLR4, but not RAGE. Bargrams represent the Ca²⁺ response (PEAK and AUC) to 10 μM NMDA alone, or in the presence of 40 nM disulfide-HMGB1, in Rage ^−/− (A) and Tlr4^−/− (B) neuronal cultures as compared with WT neurons. NMDA response was unaffected by the lack of RAGE or TLR4. In Rage ^−/− neurons (A), the potentiating effect of disulfide-HMGB1 was similar to WT, while it was precluded by LPS-RS (100 μg/ml). Disulfide HMGB1 failed to potentiate NMDA-induced Ca²⁺ response in Tlr4^−/− neurons (B). Data are the mean±SEM expressed as the % of NMDA response in WT mice (n=7–16 values from of three independent experiments). **p<0.01 versus NMDA; ^†† p<0.01 versus HMGB1+NMDA by two-way ANOVA followed by Tukey's test. RAGE, receptor for advanced glycation end products; WT, wild type.

NMDA-mediated Ca²⁺ influx in neuronal cultures from Tlr4^−/− was similar to wild-type cultures (wt, peak 0.16±0.01; AUC 0.23±0.04; Tlr4^−/− peak, 0.14±0.01; AUC, 0.23±0.02). Disulfide HMGB1 failed to potentiate NMDA-induced Ca²⁺ response in Tlr4^−/− derived neurons (Fig. 4B).

Molecular pathway underlying disulfide HMGB1 potentiation of the NMDA response

Previous evidence showed that activation of the IL-1β/IL-1R1 axis in neurons potentiates NMDA-induced Ca²⁺ influx via neutral sphingomyelinase (nSmase) and Src-family kinases, leading to the phosphorylation of the NR2B subunit (64). This pathway also mediates the proconvulsive activity of IL-1β (4). Since IL-1R1 and TLR4 have a similar cytosolic Toll/IL-1 receptor (TIR) domain and interact with the adaptor protein myeloid differentiation primary response protein (MyD88) (40), we investigated whether disulfide HMGB1 exerts its action on NMDA response by activating the same post-translational pathway as IL-1β.

Preincubation of hippocampal neurons with either 25 μM 3-O-methylsphingomyelin (3-O-MS), a selective nSmase inhibitor (57), or 10 μM 4-amino-5-(4-chlorophenyl)-7-(t-butyl) pyrazolo [3,4-d] pyrimidine (PP2), a selective antagonist of Src-family kinases (19), prevented the enhancement of the NMDA-induced Ca²⁺ response caused by 40 nM disulfide HMGB1 (p<0.05 by two-way ANOVA, Fig. 5A, B). Both inhibitors given alone were ineffective on NMDA-induced Ca²⁺ influx. PP3, the inactive PP2 analog (23), was ineffective per se, and did not affect disulfide HMGB1 action (Fig. 5B).

FIG. 5.

Involvement of nSmase, Src kinase, and NR2B phosphorylation axis in disulfide HMGB1 potentiation of the NMDA response. Bargarms represent the Ca²⁺ response (PEAK and AUC) to 10 μM NMDA+40 nM disulfide HMGB1 versus NMDA alone in the presence of 25 μM 3-O-MS (A), an inhibitor of nSmase, or 10 μM PP2 (B), an inhibitor of Src family of protein kinases. These inhibitors were ineffective when probed on NMDA alone, while both prevented the potentiating effect of disulfide HMGB1 of the NMDA response when given 30 min before, and during neuron exposure to HMGB1+NMDA. Data are the mean±SEM expressed as % of NMDA response (n=6–15 values from three independent experiments). *p<0.05, **p<0.01 versus NMDA; ^† p<0.05, and ^†† p<0.01 versus HMGB1+NMDA by two-way ANOVA followed by Tukey's test. Bargrams in (C) show the fraction of the Tyr¹⁴⁷² phosphorylated form of NR2B (p-NR2B) relative to total NR2B levels, in hippocampal homogenate of neuronal cultures. Neurons were treated for 5 min with vehicle (control), 40 nM disulfide HMGB1, or 40 nM 3S-HMGB1. Data are the mean±SEM of the optical density values of the relevant bands [depicted in representative western blots in (D)] expressed as % of control value (n=5–7 samples from three independent experiments). **p<0.01 versus control by one-way ANOVA followed by Tukey's test. 3-O-MS, 3-O-methylsphingomyelin; NR2B, N-methyl-d-aspartate receptor subunit 2B; nSmase, neutral sphingomyelinase; PP2, 4-amino-5-(4-chlorophenyl)-7-(t-butyl) pyrazolo [3,4-d] pyrimidine.

Since Src-family kinases increase NMDAR channel-gating properties by phosphorylating the NR2B subunit (49), we assessed by Western blot the level of Tyr¹⁴⁷² NR2B phosphorylation in neurons after 5 min of incubation with either disulfide or 3S-HMGB1. NR2B Tyr¹⁴⁷² phosphorylation was increased by about 55% by disulfide HMGB1 (Fig. 5C, p<0.01 by one-way ANOVA), while 3S-HMGB1 was ineffective. The total level of NR2B was not changed by the treatments (Fig. 5D).

The mechanism we describe involves short-range interactions between TLR4, its interactors and downstream effectors, and the NMDAR as a target. We expected, therefore, that TLR4 and the target NMDAR would colocalize in the same neuronal structures. Double immunostaining showed a co-localization of TLR4 with N-methyl-d-aspartate receptor subunit 1 (NR1)-NR2B (Fig. 6A, B) subunits both in the cell body and in the dendritic tree of pyramidal neurons.

FIG. 6.

Colocalization of NMDA and TLR4 receptors in hippocampal neurons. Representative immunofluorescence photomicrographs showing the co-localization of NR1 (A) and NR2B (B) (red) with TLR4 (green) in hippocampal neurons (merge in yellow). Note the colocalization signal in both cell bodies and neuronal processes. Scale bar: 25 μm. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Effect of disulfide HMGB1 on NMDA-induced excitotoxicity

To evaluate the consequences of the disulfide HMGB1-TLR4 signaling on NMDAR function, we measured neuronal death. Phase-contrast microscopy of control and disulfide HMGB1-treated neurons showed viable cells characterized by bright compact somata of round-oval shape, with intact processes (Fig. 7A). Neurons exposed to 10 μM NMDA showed signs of neurodegeneration such as loss of brightness, somatic swelling, and vacuolation (highlighted in Fig. 7A). These features were exarcerbated in cultures pretreated with disulfide HMGB1, showing also loss of membrane integrity and neurite disintegration (highlighted in Fig. 7A). Morphological assessment of cell degeneration was confirmed by quantification of lactate dehydrogenase (LDH) release in the medium (Fig. 7B): NMDA increased LDH release by 60% (p<0.01 by two-way ANOVA), and this effect was doubled by co-incubation with disulfide HMGB1 (p<0.05 by two-way ANOVA). Preincubation of hippocampal neurons with 100 μg/ml LPS-RS, a selective TLR4 antagonist, prevented disulfide HMGB1 enhancement of NMDA-induced cell death (Fig. 7A, B; p<0.01 by two-way ANOVA). LPS-RS given alone was ineffective on NMDA-induced cell death (Fig. 7A, B).

FIG. 7.

Effect of 40 n M disulfide HMGB1 on NMDA-induced neuronal excitotoxicity. (A) Shows representative phase-contrast photomicrographs of hippocampal neuronal cultures 24 h after exposure to vehicle (Mg-free Krebs), 40 nM disulfide HMGB1 (5 min), NMDA (2.15 min) alone, 40 nM disulfide-HMGB1 (5 min)+NMDA (2.15 min), 100 μg/ml LPS-RS (5 min)+NMDA (2.15 min), or to LPS-RS (5 min)+disulfide HMGB1 (5 min)+NMDA (2.15 min). The neurotoxic effect of NMDA versus control cultures was significantly exacerbated by disulfide-HMGB1, which per se was ineffective. Arrow heads indicate neuronal somatic vacuolation. Arrow indicates neurite disintegration. Disulfide-HMGB1-mediated potentiation on NMDA-induced excitotoxicity was blocked by LPS-RS. Scale bar: 25 μm. (B) Shows quantification of cell loss as assessed by LDH release in cell medium 24 h after transient drug incubation. Data are the mean±SEM expressed as % of control (n=8–26 values from seven independent experiments). **p<0.01 versus Krebs; ^† p<0.05 versus NMDA; ^‡ p<0.01 versus disulfide-HMGB1+NMDA by two-way ANOVA followed by Tukey's test. LDH, lactate dehydrogenase.

Proictogenic effect of disulfide HMGB1

When we first demonstrated that HMGB1 increased chemoconvulsant-induced seizures in mice (37), the distinction between all-thiol and disulfide HMGB1 was not appreciated, nor was the possibility that all-thiol HMGB1 could be readily oxidized in vivo. To elucidate the impact of the redox state of HMGB1 on seizures (Fig. 8A), we used 3S-HMGB1 (61), which cannot be oxidized. The intrahippocampal injection in mice of 5.5 μg all-thiol HMGB1 increased the frequency of kainic acid (KA)-induced seizures (Fig. 8B; p<0.05 by one-way ANOVA) and the total time spent in seizures by about twofold (Fig. 8C; p<0.01). An intrahippocampal injection of 5.5 μg disulfide HMGB1, 15 min before KA, induced similar proictogenic effects by increasing seizure number and duration (Fig. 8B, C; p<0.05 by one-way ANOVA), whereas an injection of 5.5 μg 3S-HMGB1 was ineffective on these parameters (Fig. 8B, C). The sample of 3S-HMGB1 we used was fully bioactive as a chemoattractant for mouse 3T3 fibroblasts (data not shown).

FIG. 8.

Effect of disulfide, all-thiol, and 3S-HMGB1 on seizures. (A) Shows a representative electro-encephalo-graphic (EEG) tracings from a mouse injected unilaterally in the hippocampus with KA. Baseline EEG activity was recorded in the left (LHP) and right (RHP) hippocampi before KA injection. Tracing in (A) (KA) depicts representative EEG seizures detected during 120 min recording after KA injection. Bargrams in (B) and (C) represent the effect of disulfide HMGB1, all-thiol HMGB1 and 3S-HMGB1 on the number of seizures (B) and time spent in seizures (C) induced by KA. Mice were intrahippocampally injected with 1 μl phosphate-buffered saline (Control) or the different forms of HMGB1 (5 μg/μl) 15 min before KA (7 ng/0.5 μl) injection in the LHP. Data are the mean±SEM (n=8–9). *p<0.05, **p<0.01 versus control; ^† p<0.05, ^†† p<0.01 versus all-thiol and disulfide HMGB1 by one-way ANOVA followed by Tukey's test. KA, kainic acid.

We suggest, therefore, that all-thiol HMGB1 is oxidized on intracerebral injection, thus acquiring the same proictogenic activity as disulfide HMGB1.

Discussion

Our study reports the novel finding that extracellular HMGB1 in its disulfide form enhances the Ca²⁺ influx evoked in neuronal cell body by NMDAR stimulation. The lack of this effect when HMGB1 is in nonoxidizable form underscores the key role of the redox state of this molecule for determining its neuronal action. The functional interaction between HMGB1 and NMDAR is mediated by the activation of TLR4, while RAGE signaling is not implicated. The TLR4-dependent molecular signaling activated by HMGB1 involves nSmase and Src kinases. Notably, the same signaling is induced in neurons by IL-1β activating IL-1R type 1 (IL-1R1) and similarly, underlies the cytokine potentiation of NMDA-mediated Ca²⁺ response (64). The common activation by both IL-1β and HMGB1 of a fast post-translational pathway in neurons which affects NMDAR function is compatible with the evidence that both TLR4 and IL-1R1 (64) are colocalized with NMDAR containing the NR2B subunit in neuronal cell body and dendrites. Moreover, IL-1R1 and TLR4 are a part of the same receptor superfamily sharing a common cytosolic TIR domain and the adapter protein MyD88, both of which are instrumental for intracellular signaling activation (40).

In analogy with IL-1β, we propose that the phosphorylation of the NR2B subunit is the final molecular target of the activation of the HMGB1-TLR4 signaling, which leads to the potentiation of NMDA-evoked neuronal Ca²⁺ influx. The following evidence supports this conclusion: (i) the disulfide, but not 3S-HMGB1, promotes both NR2B phosphorylation and NMDA-induced Ca²⁺ influx; (ii) the disulfide HMGB1 form activates Src, which is known to be involved in Tyr phosphorylation of NR2B protein and the consequent increase in magnitude and duration of NMDAR channel opening (1, 73); (iii) the disulfide HMGB1 form activates nSmase, which via ceramide production (33) alters Ca²⁺ currents in neurons through Src (4, 50, 64); and (iv) TLR4 and NR2B subunit are colocalized in the same neuronal compartments where HMGB1 potentiation of NMDA-mediated Ca²⁺ influx is measured.

Our data suggest that the activation of this rapid onset, post-translational pathway represents the key mechanism by which HMGB1 promotes neuronal excitability and seizure activity in vivo and potentiates NMDA-induced cell death (our study; 26,31,37). Notably, this was indeed the case for the pro-convulsive and pro-neurotoxic effects of IL-1β (4, 64). Concerted physiopathological actions of IL-1β and HMGB1 can be envisaged, as both these molecules are released by glia and neurons during brain injury (11, 53) or seizures (4, 37), and they cooperate to induce proinflammatory cytokines, chemokines and activate metalloproteinases (24, 52). Our evidence suggests that HMGB1 and IL-1β may also act in synergy by promoting the activation of neuronal NMDAR by endogenous ligands (e.g., glutamate, d-serine), thereby increasing neuronal excitability and excitotoxicity (Fig. 9). In seizure models, the pro-excitatory HMGB1 and IL-1β effects are associated with the phosphorylation in vivo of the NR2B subunit, and they are prevented by ifenprodil, a drug that blocks NR2B-containing NMDAR (4, 37).

FIG. 9.

Simplified representation of molecular and functional interaction between HMGB1-TLR4-NMDAR after brain injury. Brain injuries lead to the activation of neurons and glia inducing all-thiol HMGB1 nuclear-to-cytoplasmatic translocation and its cellular release. The same cell types also release glutamate. In the extracellular space, the presence of ROS and free radicals favor all-thiol HMGB1 oxidation into disulfide HMGB1 that interacts with TLR4 expressed by neurons and glial cells. HMGB1-TLR4 signaling activates a sphingomyelinase- and Src-kinases-dependent pathway in neurons, leading to the phosphorylation of the NR2B subunit and the potentiation of NMDA-mediated Ca²⁺ influx. The activation of this pathway may represent the key mechanism by which HMGB1 promotes neuronal excitability and excitotoxity. Similar effects are also induced in neurons by IL-1β, released in concert with HMGB1 by glial cells during brain injury. Disulfide HMGB1 also contributes to maintain tissue inflammation by inducing cytokine release from glia, thus establishing a vicious circle of pathologic events. IL-1β, interleukin-1 beta; ROS, reactive oxygen species. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

The redox state of HMGB1 determines its pleiotropic effects (61, 62, 69). We previously showed that convulsive stimuli cause HMGB1 nuclear-to-cytoplasmatic translocation in neurons and glia, and its cellular release (37). This phenomenon has been also described during brain ischemia (14, 44) or traumatic brain injury (18, 41). In these studies, released HMGB1 has pro-excitatory (37) and pro-neurotoxic effects (7, 14, 21, 28, 31, 41, 53, 72) as shown by pharmacological and genetic interventions in the animal models. Notably, brain injury, seizures, and the associated tissue inflammation induce the rapid production of free radicals and ROS (56, 65), thus favoring the oxidation of released HMGB1 and the consequent TLR4-dependent augmentation of neuronal NMDAR function. Disulfide HMGB1 has, indeed, been measured in experimental models of hepatic inflammation or muscle injury, showing that HMGB1 released in a permissive environment can change its redox state from a reduced to an oxidized form (62, 69). We suggest, therefore, that all-thiol HMGB1 is converted to disulfide HMGB1 in the hippocampus of mice undergoing seizures. Unfortunately, this hypothesis is difficult to prove rigorously, because the ex vivo manipulation of hippocampi in order to measure extracellular/released HMGB1 requires constant oxygenation to maintain cell viability, which would convert all-thiol HMGB1 to disulfide HMGB1 in any case.

We provide evidence that HMGB1-TLR4 functional interaction with NMDAR promotes cell death and seizures; therefore, it could critically contribute to the excitatory and neurotoxic role of HMGB1 described in brain injury models (30, 31, 44). The contribution of HMGB1 to CNS pathology may be amplified by the induction of cytokines and ROS release from immunocompetent cells and glia (17, 44, 68, 69).

There is evidence of increased neuronal HMGB1 release after incubation with NMDA at doses 2–10 higher than the one we used, and for prolonged incubation times ranging from 30 min to 24 h. These experiments were done in mixed cell cultures containing both astrocyes and neurons (31). HMGB1 levels in our neuronal cell culture medium ranged between 5 and 20 ng/ml, as assessed after 15 min and 24 h incubation in Mg²⁺-free Krebs solution. These concentrations were not significantly modified by 2.15 min exposure to NMDA followed by drug washout (data not shown). In accordance with our in vitro evidence, we demonstrated that HMGB1 in vivo is massively released by glia during seizures associated with cell loss (37), whereas glial cells are virtually absent in our culture conditions. We can, therefore, expect that glia contributes to the substantial concentrations of HMGB1 reached in vivo in the extracellular milieu in pathologic conditions. In this regard, HMGB1 brain levels have been reported to reach 2.5 μg/mg tissue after in vivo brain ischemia (66). These concentrations are in the range of those we used in our study.

HMGB1 potentiation of postsynaptic NMDAR function may also have a role in the cognitive deficits described in animals after systemic (8) or intracerebral HMGB1 administration in mice (39). Thus, a relatively high dose of HMGB1 induces loss of dendritic spines in pyramidal neurons that are associated with memory dysfunction (8). Notably, NR2B phosphorylation has been shown to mediate dendritic spine loss induced by amyloid-β-oligomers in hippocampal cultures (58).

A recent study indicated that HMGB1 can interact directly with presynaptic NMDAR without the concomitant activation of TLR4 or RAGE (42). The mechanism we describe here is different: (i) We monitored Ca²⁺ influx induced by NMDA in cell somata, therefore our mechanism involves postsynaptic NMDAR. Presynaptic NMDAR-mediated effects can be excluded, as our experiments were run in the presence of tetrodotoxin (TTX), which blocks synaptic transmission; (ii) the molecular interaction between HMGB1 and presynaptic NMDAR facilitates receptor activation at sub-stimulatory amounts of agonist (42); in our studies, HMGB1 potentiates the postsynaptic NMDA response only when NMDAR is activated by an effective agonist concentration. This likely depends on the ability of HMGB1 to amplify the magnitude and duration of the NMDA-activated Ca²⁺ permeable channel (1, 73) by inducing NR2B phosphorylation; (iii) the physical interaction between HMGB1 and NMDAR was observed at picomolar HMGB1 concentrations and promoted cell motility, neurite outgrowth, cell differentiation, and glutamate release from presynaptic terminals, which are phenomena with putative physiological effects during brain development, cell repair/regeneration, and long-term potentiation (13, 45, 46, 60). We instead showed the activation of a pathological chain of events triggered by nanomolar HMGB1 concentrations, which are similar to those measured in rodent brain after damage or seizures (37, 66). The combination of our data and those already published, therefore, indicates that the functional consequences of extracellular HMGB1 on neurons depend on its redox state, extracellular concentration, and the neuronal compartment where HMGB1–NMDAR interactions take place (i.e., cell somata vs. presynaptic terminals).

In conclusion, our findings provide novel evidence of a TLR4-mediated interaction between extracellular HMGB1 and neuronal NR2B containing NMDAR, leading to pathologic outcomes. Agents interfering with this HMGB1 action, such as BoxA or TLR4 antagonists, may be, therefore, considered for developing pharmacological interventions (20, 71) to prevent the excessive NMDAR stimulation leading to hyperexcitability and cell damage in various acute and chronic CNS disorders (15).

Materials and Methods

Experimental animals

We used C57BL6N wild-type (Charles River), Rage⁻ ^/−, and Tlr4⁻ ^/− 18 day-old mouse fetuses for neuronal cultures and male C57BL6N mice (60 day-old, 25 g) for EEG recording. The generation of the Rage ^−/− mouse line has been previously described in detail (3, 35). Tlr4 ^−/− mice were obtained from The Jackson Laboratories (Strain Name: C57BL/10ScNJ). They have a deletion of the Tlr4 gene, resulting in the absence of both mRNA and protein; thus, these mice do not respond to lipopolysaccharide (LPS) stimulation. The phenotypic characterization of these mice is reported in detail in the vendor Web site (C57BL/10ScNJ, Stock Number 003752; http://jaxmice.jax.org/strain/003752.html).

Procedures involving animals and their care were conducted in accordance with the ethically approved institutional guidelines that are in compliance with national and international laws and policies (EEC Council Directive 86/609, OJ L 358, 1, Dec.12, 1987; Guide for the Care and Use of Laboratory Animals, U.S. National Research Council, 1996).

Primary hippocampal neuronal cultures

Neuronal cultures were prepared from C57BL6N wild-type Rage⁻ ^/− and Tlr4⁻ ^/− 18 day-old mouse fetuses as previously described (9). Hippocampi were isolated in cold Hank's Buffered Salt Solution (HBSS; Life Technologies Invitrogen), then incubated 20 min at 37°C in HBSS containing 0.1% trypsin (Life Technologies Invitrogen). The digestion was then blocked with 10% heat-inactivated fetal bovine serum (Life Technologies Invitrogen) and 0.05% DNAse (Sigma) in Neurobasal media (Life Technologies Invitrogen). The hippocampi were mechanically dissociated by repeated passage through a constricted pasteur pipette. The cells were resuspended in Dulbecco's Modified Eagle Medium (Life Technologies Invitrogen) containing 10% horse serum, and they were plated on eight-well microslides (Ibidi) coated with 25 μg/ml poly-d-lysine (Sigma) and 1 μg/ml laminin; 3.5×10⁴ cells per well were used for Ca²⁺ imaging experiments and immunocytochemistry. For neuronal cell death detection, 2.5×10⁵ neuronal cells per well were plated on 12-well plates; for western blot analysis, we used 3.5×10⁵ neuronal cells per well. After 2–3 h at 37°C to allow cell attachment to the substrate, the medium was replaced with neurobasal medium supplemented with 2% B27 and 2 mM glutamine (Life Technologies Invitrogen). Three days after plating, 1 μM cytosine arabinoside was added for 48 h to inhibit glia proliferation. Double labeling with antibodies against Neu-N (1:1000; Merck Millipore) and glial fibrillary acidic protein (1:500; Dako) showed that >90% of the cells were neurons. Cultures were kept at 37°C and 5% CO₂ and used for experiments at 11–13 days in vitro.

Single-cell Ca²⁺ imaging

Cultured hippocampal neurons plated into eight-well microslides were loaded for 30 min at 37°C with 10 μM fura-2AM form (Molecular Probes) in Krebs–Ringer–Hepes (KRH) buffer (NaCl 125.0 mM, KCl 5.0 mM, MgSO₄ 1.2 mM, CaCl₂ 2.0 mM, glucose 10.0 mM Hepes 25.0 mM, and pH 7.4) supplemented with 1% bovine serum albumin (BSA) (63). After washing out the dye-containing medium, cells were imaged on the stage of an inverted Olympus IX81 microscope equipped with a Ca²⁺ imaging unit (Cell^R Olympus; 51). Image sequences were acquired for 3 min by alternate excitation at 340 and 380 nm, and measurement of emission at 510 nm every 235 ms (9 ms exposure time, 2×2 binning) with 40×fluorescence objective and a digital CCD camera. Fura-2AM fluorescence intensity, which reflects the concentration of intracellular Ca²⁺, was measured in all single neuronal somata present in the recorded field and was expressed as DeltaF/F (ΔF/F), which corresponds to (Ft−F0/F0), where F0 is the initial fluorescence value and Ft is the fluorescence value at the time t of measurement. All imaging experiments were carried out at 37°C in 5% CO₂ Mg²⁺-free KRH containing 0.5 μM TTX to prevent synaptic activity.

The response to drug application was calculated as the peak amplitude defined as the maximum Ca²⁺ rise and the AUC obtained from the integral of the area of the response for each cell (Prism Graphpad Software) as an index of the total Ca²⁺ entry (34).

Drug application

NMDA (Tocris) was dissolved in Mg²⁺-free KRH and applied at 1–500 μM to neuronal hippocampal cultures for 2.15 min as a bolus application after 45 s of baseline fluorescence recording (64). MK-801 (10 μM; Tocris), a use-dependent open channel NMDAR antagonist (67), was incubated for 5 min before 10 μM NMDA, and along with NMDA for 2.15 min (64).

Disulfide HMGB1 (HMGBiotech) was expressed in Escherichia coli and purified using a continuous descending gradient of ammonium sulfate (HiLoad 26/10 Phenyl Sepharose High Performance column connected to an Akta Purifier FPLC system; GE Healthcare) followed by ionic-strength chromatography (Hi-trap Q column; GE Healthcare). The purity and integrity of purified HMGB1 was verified by Coomassie blue staining after sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Contaminating LPS was removed from protein preparations by Triton X-114 extraction. Disulfide HMGB1 activity was measured by testing its ability to stimulate cytokine production in human macrophages.

Nonoxidizable HMGB1 (3S-HMGB1) was generated using the QuikChange XL Site-Directed Mutagenesis kit according to the manufacturer's instructions (Stratagene; Agilent Technologies), checked by sequencing, expressed, and purified as Cytokine-HMGB1. The activity of 3S-HMGB1 was measured by its ability to induce migration of mouse 3T3 fibroblasts in migration assay.

Full-length, LPS-free recombinant HMGB1 (all-thiol) is the unmodified protein, with no tags or additional amino acids. It is produced in E. coli from an expression plasmid coding for the rat protein, and it was concentrated for an injection in the brain using Microcon 10000 MWCO filters (Millipore). In in vitro experiments, disulfide, all-thiol, or 3S-HMGB1 (0.004–400 nM) was added 5 min before, and during, 10 μM NMDA stimulation.

In in vivo experiments, disulfide, all-thiol or 3S-HMGB1 (5.5 μg/μl) was unilaterally injected 15 min before KA into the septal pole of the hippocampus.

BoxA (HMGBiotech) corresponds to one of the highly conserved DNA-binding domains of HMGB1 protein, aa 1–89. It functions as endogenous HMGB1 competitor (70). BoxA is extensively tested by HMGBiotech to exclude LPS contamination for HMGB1. BoxA (100 nM) was incubated 5 min before HMGB1, and for the following incubation times as reported in the respective figures. This was the dose effective in suppressing cortical neuronal cell death induced by conditioned media from NMDA-treated cultures (31).

LPS-RS (100 μg/ml; InvivoGen), a selective TLR4 antagonist (10), was added 5 min before HMGB1. One-hundred-fold excess relative to HMGB1 was previously used to inhibit LPS-induced intracellular Ca²⁺ increase in trigeminal sensory neurons (12).

3-O-MS (25 μM; BioMol Research Laboratories, Inc.), a selective nSmase inhibitor (57, 74), was added 30 min before HMGB1. This dose was shown to inhibit the IL-1β-induced IL-6 synthesis in primary neurons (57).

Ten micromolar PP2 (Calbiochem), an Src-family of tyrosine kinase inhibitor (19), was added 30 min before HMGB1 and for the following incubation times as reported in the respective figures. This dose was shown to block IL-1β-mediated hyperpolarization of warm-sensitive hypothalamic neurons (50). Control cultures were treated with 10 μM PP3, the ineffective analog of PP2.

Immunofluorescence labeling and confocal imaging

Hippocampal neurons were fixed in 4% paraformaldehyde for 30 min, then washed thrice with phosphate-buffered saline (PBS). Cells were permeabilized at 4°C in PBS containing 0.1% Triton X-100 (Sigma) for 3 min followed by 15 min incubation in 10% BSA in PBS to inhibit nonspecific antibody binding sites. Then, neurons were incubated overnight with primary antibodies against NR1 subunit (1:200; Merck Millipore), NR2B subunit (1:50; Santa Cruz Biotechnology, Inc.), and TLR4 (1:200; Abcam, Inc.) at 4°C in 0.2% BSA in PBS. After three washing steps, sections were incubated in PBS containing anti-mouse (for NR1), anti-goat (for NR2B), or anti-rabbit (for TLR4) secondary antibody conjugated to Alexa488 or Alexa546 (1:500; Molecular Probes). Double immunostaining was examined with an Olympus Fluorview laser scanning confocal microscope (microscope BX61 and confocal system FV500) using dual excitation of 488 nm (Laser Ar) and 546 nm (Laser He–Ne green) for Alexa488 and Alexa546, respectively. The emission of fluorescent probes was collected on separate detectors. To eliminate the possibility of bleed-through between channels, the sections were scanned in a sequential mode. The specificity of the TLR4 antibody was double checked using neurons from Tlr4 ^−/− mice. In these neurons lacking Tlr4, no staining was observed. As additional control condition, omission of the primary antibody against NR1, NR2B, or Tlr4 abrogated the immunohistochemical signal.

Western blots

Hippocampal neuronal cultures were treated for 5 min with 40 nM disulfide or 3S-HMGB1. After two washes with ice-cold PBS, the cells were harvested into ice-cold solubilization buffer containing 20 mM Tris acetate, 270 mM sucrose, 0.1% deoxycholate, 1 mM ethylenediaminetetraaceticacid, 1 mM ethylene glycol tetraacetic acid, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na ortovanadate, 50 mM NaF, 5 mM Na pirophosphate, 10 mM Na-β-glicerophosphate, 1 mM dl-dithio-threitol, 1% Triton X-100 in the presence of a complete set of protease (Complete; Boehringer Mannheim), and phosphatase inhibitors (Sigma). Total proteins (30 μg per lane; Bio-Rad Protein Assay) were separated using 8% acrylamide SDS-PAGE, and each sample was run in duplicate. Proteins were transferred to Hybond nitrocellulose membranes by electroblotting. For immunoblotting, we used anti-pTyr¹⁴⁷²-NR2B (1:1000; Thermo Scientific) rabbit polyclonal antibodies. Blots used to measure pTyr¹⁴⁷²-NR2B were stripped, then re-probed to measure total levels of NR2B as previously described (16). Total NR2B levels were assessed using rabbit polyclonal anti-NR2B antibody (1:500; Invitrogen). Immunoreactivity was visualized with enhanced chemiluminescence (ECL) using peroxidase-conjugated goat anti-rabbit (1:1000; Sigma) immunoglobulin as secondary antibodies. Densitometric analysis of immunoblots was done to quantify the changes in protein levels (Quantity One Software; Bio-Rad Laboratories) using film exposures with maximal signals below the photographic saturation point. Optical density values in each sample were normalized using the corresponding amount of β-actin (1:15,000; Santa Cruz Biotechnology).

Cellular viability assay

Hippocampal neurons were exposed to vehicle (Mg²⁺-free KRH), 40 nM disulfide HMGB1 alone for 5 min, 10 μM NMDA alone for 2.15 min, or LPS-RS (100 μg/ml)±disulfide HMGB1 for 5 min followed by co-incubation with 10 μM NMDA. At the end of incubation, all drugs were removed and neurons were incubated in fresh culture medium for 24 h. Neuronal cell damage was assessed by measuring LDH released in the extracellular medium (LDH detection kit; Roche). LDH release is a well-established index of cell injury (32, 36). Fifty microliter of culture medium were combined with 50 μl of NAD⁺ and tetrazolium salt solution for 30 min; then, the absorbance was detected at 492 nm with a spectrophotometer (ELISA reader; Infinite M200). The results were expressed as % of LDH values in control culture medium. At the end of the experiments, cell cultures were fixed in 4% paraformaldehyde for 30 min to assess cell viability morphologically by phase-contrast microscopy.

Mouse model of seizures

Male C57BL6N mice (60 day-old, 25 g) mice were surgically implanted with an injection guide cannula and recording electrodes under general gas anesthesia (1.4% isoflurane in a mixture of 70% N₂O–30% O₂) and stereotaxic guidance (37). Two nichrome-insulated bipolar depth electrodes (60 μm OD) were implanted bilaterally into the dorsal hippocampus (from bregma [mm]: nose bar 0; anteroposterior—1.7, lateral 1.5 and 1.9 below dura mater). A 23-gauge cannula was unilaterally positioned on top of the dura mater and glued to one of the depth electrodes for the intrahippocampal infusion of KA. The electrodes were connected to a multipin socket and, along with the cannula of injection, secured to the skull by acrylic dental cement. The correct position of the electrodes and injection needle was evaluated by histological analysis of brain tissue at the end of the experiments.

An intrahippocampal injection of KA in freely moving mice was adiministered 7 days after surgery as previously described (37). KA (7 ng in 0.5 μl; Sigma) was dissolved in 0.1 M PBS, pH 7.4 and injected unilaterally in the dorsal hippocampus by using a needle protruding 1.9 mm from the bottom of the guide cannula. This dose of KA was proved to induce EEG ictal episodes in the hippocampus in 100% of mice without mortality (37).

Acute seizures assessment and quantification

EEG seizures induced by an intrahippocampal injection of KA in mice have been extensively described earlier (26, 37). Briefly, a 30 min baseline EEG recording was done before KA injection to assess the spontaneous activity pattern, and for 180 min after kainate injection. Kainate was injected over 1 min (CMA/100 pump) using a 30-gauge injection needle connected to a 10.0 μl Hamilton microsyringe via PE20 tubing; the needle was left in place for one additional min to avoid backflow through the cannula. Ictal episodes are characterized by high frequency (7–10 Hz) and/or multispike complexes and/or high-voltage (700 μV–1.0 mV) synchronized spikes simultaneously occurring in the injected and contralateral hippocampi (Fig. 8A). EEG activity was monitored using the GRASS 79D EEG Recording System. The signal was digitalized with a PowerLab 16/S data acquisition system and analyzed with LabChart 7 software (ADInstruments Pty Ltd.). The EEG recording of each animal was analyzed visually to detect any activity different from baseline. Seizures were quantified by their total number and duration (by summing up the duration of every ictal episode during the EEG recording period). Seizures occurred with an average latency of about 10 min from KA injection, then recurred for about 90 min from their onset, and were associated with motor arrest of the mice. EEG analysis was done by two independent investigators who were blinded to the treatment, who visually reviewed all the EEG tracings.

Statistical analysis

Data represented the means±standard error of the mean (n=number of individual wells). The effects of treatments were analyzed by one-way ANOVA followed by Tukey's test (NMDA dose–response, western blot analysis, and EEG seizures) or by two-way ANOVA followed by Tukey's test or Student t test. A significance level of 95% (p<0.05) was accepted. Statistical analysis was done using absolute values.

Footnotes

Acknowledgments

This work was supported by Cariplo Foundation grant 2009-2426 (A.V. and M.E.B.), Fondazione Monzino (A.V.), and Ministero della salute (RF-2009-1506142 to A.V. and M.E.B.). The authors are grateful to Pietro Veglianese, Assunta Senatore (Mario Negri Institute), and Alessandro Preti (HMGBiotech) for their contribution to a part of these experiments.

Author Disclosure Statement

The authors declare no direct financial interest. However, M.E.B. is founder and part-owner of HMGBiotech http://hmgbiotech.com, a company that provides goods and services related to HMGB proteins.

Abbreviations Used

References

Ali

and Salter

. Nmda receptor regulation by src kinase signalling in excitatory synaptic transmission and plasticity. Curr Opin Neurobiol, 11: 336–342, 2001.

Andersson

and Rauvala

. Introduction: Hmgb1 in inflammation and innate immunity. J Intern Med, 270: 296–300, 2011.

Andrassy

, Volz

, Igwe

, Funke

, Eichberger

, Kaya

, Buss

, Autschbach

, Pleger

, Lukic

, Bea

, Hardt

, Humpert

, Bianchi

, Mairbaurl

, Nawroth

, Remppis

, Katus

, and Bierhaus

. High-mobility group box-1 in ischemia-reperfusion injury of the heart. Circulation, 117: 3216–3226, 2008.

Balosso

, Maroso

, Sanchez-Alavez

, Ravizza

, Frasca

, Bartfai

, and Vezzani

. A novel non-transcriptional pathway mediates the proconvulsive effects of interleukin-1beta. Brain, 131: 3256–3265, 2008.

Bianchi

. Damps, pamps and alarmins: all we need to know about danger. J Leukoc Biol, 81: 1–5, 2007.

Bianchi

and Manfredi

. High-mobility group box 1 (hmgb1) protein at the crossroads between innate and adaptive immunity. Immunol Rev, 220: 35–46, 2007.

Caso

, Pradillo

, Hurtado

, Leza

, Moro

, and Lizasoain

. Toll-like receptor 4 is involved in subacute stress-induced neuroinflammation and in the worsening of experimental stroke. Stroke, 39: 1314–1320, 2008.

Chavan

, Huerta

, Robbiati

, Valdes-Ferrer

, Ochani

, Dancho

, Frankfurt

, Volpe

, Tracey

, and Diamond

. Hmgb1 mediates cognitive impairment in sepsis survivors. Mol Med, 18: 930–937, 2012.

Chiarugi

. Characterization of the molecular events following impairment of nf-kappab-driven transcription in neurons. Brain Res Mol Brain Res, 109: 179–188, 2002.

10.

Coats

, Pham

, Bainbridge

, Reife

, and Darveau

. Md-2 mediates the ability of tetra-acylated and penta-acylated lipopolysaccharides to antagonize Escherichia coli lipopolysaccharide at the tlr4 signaling complex. J Immunol, 175: 4490–4498, 2005.

11.

Crews

, Qin

, Sheedy

, Vetreno

, and Zou

. High mobility group box 1/toll-like receptor danger signaling increases brain neuroimmune activation in alcohol dependence. Biol Psychiatry, 73: 602–612, 2012.

12.

Diogenes

, Ferraz

, Akopian

, Henry

, and Hargreaves

. Lps sensitizes trpv1 via activation of tlr4 in trigeminal sensory neurons. J Dent Res, 90: 759–764, 2011.

13.

Fang

, Schachner

, and Shen

. Hmgb1 in development and diseases of the central nervous system. Mol Neurobiol, 45: 499–506, 2012.

14.

Faraco

, Fossati

, Bianchi

, Patrone

, Pedrazzi

, Sparatore

, Moroni

, and Chiarugi

. High mobility group box 1 protein is released by neural cells upon different stresses and worsens ischemic neurodegeneration in vitro and in vivo . J Neurochem, 103: 590–603, 2007.

15.

Forder

and Tymianski

. Postsynaptic mechanisms of excitotoxicity: involvement of postsynaptic density proteins, radicals, and oxidant molecules. Neuroscience, 158: 293–300, 2009.

16.

Frasca

, Aalbers

, Frigerio

, Fiordaliso

, Salio

, Gobbi

, Cagnotto

, Gardoni

, Battaglia

, Hoogland

, Di Luca

, and Vezzani

. Misplaced nmda receptors in epileptogenesis contribute to excitotoxicity. Neurobiol Dis, 43: 507–515, 2011.

17.

Gao

, Zhou

, Zhang

, Wilson

, Kam

, and Hong

. Hmgb1 acts on microglia mac1 to mediate chronic neuroinflammation that drives progressive neurodegeneration. J Neurosci, 31: 1081–1092, 2011.

18.

Gao

, Yuan

, Yang

, Dai

, Wang

, Peng

, Shao

, Jin

, and Fu

. Expression of hmgb1 and rage in rat and human brains after traumatic brain injury. J Trauma Acute Care Surg, 72: 643–649, 2012.

19.

Hanke

, Gardner

, Dow

, Changelian

, Brissette

, Weringer

, Pollok

, and Connelly

. Discovery of a novel, potent, and src family-selective tyrosine kinase inhibitor. Study of lck- and fynt-dependent t cell activation. J Biol Chem, 271: 695–701, 1996.

20.

Hanke

and Kielian

. Toll-like receptors in health and disease in the brain: mechanisms and therapeutic potential. Clin Sci (Lond), 121: 367–387, 2011.

21.

Hayakawa

, Mishima

, Irie

, Hazekawa

, Mishima

, Fujioka

, Orito

, Egashira

, Katsurabayashi

, Takasaki

, Iwasaki

, and Fujiwara

. Cannabidiol prevents a post-ischemic injury progressively induced by cerebral ischemia via a high-mobility group box1-inhibiting mechanism. Neuropharmacology, 55: 1280–1286, 2008.

22.

Hoppe

, Talcott

, Bhattacharya

, Crabb

, and Sears

. Molecular basis for the redox control of nuclear transport of the structural chromatin protein hmgb1. Exp Cell Res, 312: 3526–3538, 2006.

23.

Hou

, Liu

, and Zhang

. Pp2, a potent inhibitor of src family kinases, protects against hippocampal ca1 pyramidal cell death after transient global brain ischemia. Neurosci Lett, 420: 235–239, 2007.

24.

Hreggvidsdottir

, Ostberg

, Wahamaa

, Schierbeck

, Aveberger

, Klevenvall

, Palmblad

, Ottosson

, Andersson

, and Harris

. The alarmin hmgb1 acts in synergy with endogenous and exogenous danger signals to promote inflammation. J Leukoc Biol, 86: 655–662, 2009.

25.

Huang

, Smith

, Ibanez-Sandoval

, Sims

, and Friedman

. Neuron-specific effects of interleukin-1beta are mediated by a novel isoform of the il-1 receptor accessory protein. J Neurosci, 31: 18048–18059, 2011.

26.

Iori

, Maroso

, Rizzi

, Iyer

, Vertemara

, Carli

, Agresti

, Antonelli

, Bianchi

, Aronica

, Ravizza

, and Vezzani

. Receptor for advanced glycation endproducts is upregulated in temporal lobe epilepsy and contributes to experimental seizures. Neurobiol Dis, 58: 102–114, 2013.

27.

Kalia

, Kalia

, and Salter

. Nmda receptors in clinical neurology: excitatory times ahead. Lancet Neurol, 7: 742–755, 2008.

28.

Kim

, Lim

, Kim

, Nam

, Kim

, Park

, and Lee

. Neuroprotection by biodegradable pamam ester (e-pam-r)-mediated hmgb1 sirna delivery in primary cortical cultures and in the postischemic brain. J Control Release, 142: 422–430, 2010.

29.

Kim

, Wan

, JOC

, Shaikh

, and Nicholson

. The role of receptor for advanced glycation end products (rage) in neuronal differentiation. J Neurosci Res, 90: 1136–1147, 2012.

30.

Kim

, Sig Choi

, Yu

, Nam

, Piao

, Kim

, Lee

, Han

, Park

, and Lee

. Hmgb1, a novel cytokine-like mediator linking acute neuronal death and delayed neuroinflammation in the postischemic brain. J Neurosci, 26: 6413–6421, 2006.

31.

Kim

, Lim

, Kim

, Shin

, Lee

, and Lee

. Extracellular hmgb1 released by nmda treatment confers neuronal apoptosis via rage-p38 mapk/erk signaling pathway. Neurotox Res, 20: 159–169, 2011.

32.

Koh

and Choi

. Quantitative determination of glutamate mediated cortical neuronal injury in cell culture by lactate dehydrogenase efflux assay. J Neurosci Methods, 20: 83–90, 1987.

33.

Kolesnick

and Golde

. The sphingomyelin pathway in tumor necrosis factor and interleukin-1 signaling. Cell, 77: 325–328, 1994.

34.

Koss

, Riedel

, and Platt

. Intracellular ca2+ stores modulate soccs and nmda receptors via tyrosine kinases in rat hippocampal neurons. Cell Calcium, 46: 39–48, 2009.

35.

Liliensiek

, Weigand

, Bierhaus

, Nicklas

, Kasper

, Hofer

, Plachky

, Grone

, Kurschus

, Schmidt

, Yan

, Martin

, Schleicher

, Stern

, Hammerling

, Nawroth

, and Arnold

. Receptor for advanced glycation end products (rage) regulates sepsis but not the adaptive immune response. J Clin Invest, 113: 1641–1650, 2004.

36.

Lobner

. Comparison of the ldh and mtt assays for quantifying cell death: validity for neuronal apoptosis?. J Neurosci Methods, 96: 147–152, 2000.

37.

Maroso

, Balosso

, Ravizza

, Liu

, Aronica

, Iyer

, Rossetti

, Molteni

, Casalgrandi

, Manfredi

, Bianchi

, and Vezzani

. Toll-like receptor 4 and high-mobility group box-1 are involved in ictogenesis and can be targeted to reduce seizures. Nat Med, 16: 413–419, 2010.

38.

Martin

, Bustos

, Bowe

, Bray

, and Nadler

. Autoreceptor regulation of glutamate and aspartate release from slices of the hippocampal ca1 area. J Neurochem, 56: 1647–1655, 1991.

39.

Mazarati

, Maroso

, Iori

, Vezzani

, and Carli

. High-mobility group box-1 impairs memory in mice through both toll-like receptor 4 and receptor for advanced glycation end products. Exp Neurol, 232: 143–148, 2011.

40.

O'Neill

and Bowie

. The family of five: Tir-domain-containing adaptors in toll-like receptor signalling. Nat Rev Immunol, 7: 353–364, 2007.

41.

Okuma

, Liu

, Wake

, Zhang

, Maruo

, Date

, Yoshino

, Ohtsuka

, Otani

, Tomura

, Shima

, Yamamoto

, Takahashi

, Mori

, and Nishibori

. Anti-high mobility group box-1 antibody therapy for traumatic brain injury. Ann Neurol, 72: 373–384, 2012.

42.

Pedrazzi

, Averna

, Sparatore

, Patrone

, Salamino

, Marcoli

, Maura

, Cervetto

, Frattaroli

, Pontremoli

, and Melloni

. Potentiation of nmda receptor-dependent cell responses by extracellular high mobility group box 1 protein. PLoS One, 7: e44518, 2012.

43.

Pedrazzi

, Patrone

, Passalacqua

, Ranzato

, Colamassaro

, Sparatore

, Pontremoli

, and Melloni

. Selective proinflammatory activation of astrocytes by high-mobility group box 1 protein signaling. J Immunol, 179: 8525–8532, 2007.

44.

Qiu

, Nishimura

, Wang

, Sims

, Qiu

, Savitz

, Salomone

, and Moskowitz

. Early release of hmgb-1 from neurons after the onset of brain ischemia. J Cereb Blood Flow Metab, 28: 927–938, 2008.

45.

Rauvala

and Rouhiainen

. Physiological and pathophysiological outcomes of the interactions of hmgb1 with cell surface receptors. Biochim Biophys Acta, 1799: 164–170, 2010.

46.

Rong

, Trojaborg

, Qu

, Kostov

, Yan

, Gooch

, Szabolcs

, Hays

, and Schmidt

. Antagonism of rage suppresses peripheral nerve regeneration. FASEB J, 18: 1812–1817, 2004.

47.

Rubartelli

and Lotze

. Inside, outside, upside down: damage-associated molecular-pattern molecules (damps) and redox. Trends Immunol, 28: 429–436, 2007.

48.

Sahu

, Debnath

, Takayama

, and Iwahara

. Redox properties of the a-domain of the hmgb1 protein. FEBS Lett, 582: 3973–3978, 2008.

49.

Salter

and Kalia

. Src kinases: a hub for nmda receptor regulation. Nat Rev Neurosci, 5: 317–328, 2004.

50.

Sanchez-Alavez

, Tabarean

, Behrens

, and Bartfai

. Ceramide mediates the rapid phase of febrile response to il-1beta. Proc Natl Acad Sci U S A, 103: 2904–2908, 2006.

51.

Senatore

, Colleoni

, Verderio

, Restelli

, Morini

, Condliffe

, Bertani

, Mantovani

, Canovi

, Micotti

, Forloni

, Dolphin

, Matteoli

, Gobbi

, and Chiesa

. Mutant prp suppresses glutamatergic neurotransmission in cerebellar granule neurons by impairing membrane delivery of vgcc alpha(2)delta-1 subunit. Neuron, 74: 300–313, 2012.

52.

Sha

, Zmijewski

, Xu

, and Abraham

. Hmgb1 develops enhanced proinflammatory activity by binding to cytokines. J Immunol, 180: 2531–2537, 2008.

53.

, Wang

, Zhao

, Pan

, and Mao

. Beneficial effects of ethyl pyruvate through inhibiting high-mobility group box 1 expression and tlr4/nf-kappab pathway after traumatic brain injury in the rat. Mediators Inflamm, 2011: 807142, 2011.

54.

Suarez

, Suarez

, Del Olmo

, Ruiz

, Gonzalez-Escalada

, and Solis

. Presynaptic nmda autoreceptors facilitate axon excitability: a new molecular target for the anticonvulsant gabapentin. Eur J Neurosci, 21: 197–209, 2005.

55.

Tang

, Billiar

, and Lotze

. A janus tale of two active high mobility group box 1 (hmgb1) redox states. Mol Med, 18: 1360–1362, 2012.

56.

Tang

, Kang

, Zeh

3rd , and Lotze

. High-mobility group box 1, oxidative stress, and disease. Antioxid Redox Signal, 14: 1315–1335, 2011.

57.

Tsakiri

, Kimber

, Rothwell

, and Pinteaux

. Interleukin-1-induced interleukin-6 synthesis is mediated by the neutral sphingomyelinase/src kinase pathway in neurones. Br J Pharmacol, 153: 775–783, 2008.

58.

, Nygaard

, Heiss

, Kostylev

, Stagi

, Vortmeyer

, Wisniewski

, Gunther

, and Strittmatter

. Alzheimer amyloid-beta oligomer bound to postsynaptic prion protein activates fyn to impair neurons. Nat Neurosci, 15: 1227–1235, 2012.

59.

Van Dongen

. Biology of the Nmda Receptor. Boca Raton, FL: CRC Press, 2009, p. 342.

60.

Vedder

, Smith

, Flannigan

, and McMahon

. Estradiol-induced increase in novel object recognition requires hippocampal nr2b-containing nmda receptors. Hippocampus, 23: 108–115, 2013.

61.

Venereau

, Casalgrandi

, Schiraldi

, Antoine

, Cattaneo

, De Marchis

, Liu

, Antonelli

, Preti

, Raeli

, Shams

, Yang

, Varani

, Andersson

, Tracey

, Bachi

, Uguccioni

, and Bianchi

. Mutually exclusive redox forms of hmgb1 promote cell recruitment or proinflammatory cytokine release. J Exp Med, 209: 1519–1528, 2012.

62.

Venereau

, Schiraldi

, Uguccioni

, and Bianchi

. Hmgb1 and leukocyte migration during trauma and sterile inflammation. Mol Immunol, 55: 76–82, 2013.

63.

Verderio

, Pozzi

, Pravettoni

, Inverardi

, Schenk

, Coco

, Proux-Gillardeaux

, Galli

, Rossetto

, Frassoni

, and Matteoli

. Snap-25 modulation of calcium dynamics underlies differences in gabaergic and glutamatergic responsiveness to depolarization. Neuron, 41: 599–610, 2004.

64.

Viviani

, Bartesaghi

, Gardoni

, Vezzani

, Behrens

, Bartfai

, Binaglia

, Corsini

, Di Luca

, Galli

, and Marinovich

. Interleukin-1beta enhances nmda receptor-mediated intracellular calcium increase through activation of the src family of kinases. J Neurosci, 23: 8692–8700, 2003.

65.

Waldbaum

and Patel

. Mitochondria, oxidative stress, and temporal lobe epilepsy. Epilepsy Res, 88: 23–45, 2010.

66.

Wang

, Wang

, Li

, Yang

, Xu

, Huang

, Zhang

, Gou

, Chen

, and Xiong

. Electroacupuncture pretreatment attenuates cerebral ischemic injury through alpha7 nicotinic acetylcholine receptor-mediated inhibition of high-mobility group box 1 release in rats. J Neuroinflammation, 9: 24, 2012.

67.

Wong

, Kemp

, Priestley

, Knight

, Woodruff

, and Iversen

. The anticonvulsant mk-801 is a potent n-methyl-d-aspartate antagonist. Proc Natl Acad Sci U S A, 83: 7104–7108, 1986.

68.

Yang

, Hreggvidsdottir

, Palmblad

, Wang

, Ochani

, Li

, Lu

, Chavan

, Rosas-Ballina

, Al-Abed

, Akira

, Bierhaus

, Erlandsson-Harris

, Andersson

, and Tracey

. A critical cysteine is required for hmgb1 binding to toll-like receptor 4 and activation of macrophage cytokine release. Proc Natl Acad Sci U S A, 107: 11942–11947, 2010.

69.

Yang

, Lundback

, Ottosson

, Erlandsson-Harris

, Venereau

, Bianchi

, Al-Abed

, Andersson

, Tracey

, and Antoine

. Redox modification of cysteine residues regulates the cytokine activity of high mobility group box-1 (hmgb1). Mol Med, 18: 250–259, 2012.

70.

Yang

, Ochani

, Li

, Qiang

, Tanovic

, Harris

, Susarla

, Ulloa

, Wang

, DiRaimo

, Czura

, Roth

, Warren

, Fink

, Fenton

, Andersson

, and Tracey

. Reversing established sepsis with antagonists of endogenous high-mobility group box 1. Proc Natl Acad Sci U S A, 101: 296–301, 2004.

71.

Yang

, Wang

, and Czura

, Tracey

. Hmgb1 as a cytokine and therapeutic target. J Endotoxin Res, 8: 469–472, 2002.

72.

Yang

, Xiang

, Zhou

, Zhong

, and Li

. Targeting hmgb1/tlr4 signaling as a novel approach to treatment of cerebral ischemia. Front Biosci (Schol Ed), 2: 1081–1091, 2010.

73.

, Askalan

, Keil

2nd , and Salter

. Nmda channel regulation by channel-associated protein tyrosine kinase src. Science, 275: 674–678, 1997.

74.

Zeng

, Lee

, Chen

, Hsu

, and Xu

. Amyloid-beta peptide enhances tumor necrosis factor-alpha-induced inos through neutral sphingomyelinase/ceramide pathway in oligodendrocytes. J Neurochem, 94: 703–712, 2005.

Disulfide-Containing High Mobility Group Box-1 Promotes N -Methyl- d -Aspartate Receptor Function and Excitotoxicity by Activating Toll-Like Receptor 4-Dependent Signaling in Hippocampal Neurons

Abstract

Introduction

Results

NMDAR-mediated neuronal Ca2+ influx

Effect of HMGB1 redox state on NMDA-induced Ca2+ increase

Receptor-mediated effect of disulfide HMGB1

Molecular pathway underlying disulfide HMGB1 potentiation of the NMDA response

Effect of disulfide HMGB1 on NMDA-induced excitotoxicity

Proictogenic effect of disulfide HMGB1

Discussion

Materials and Methods

Experimental animals

Primary hippocampal neuronal cultures

Single-cell Ca2+ imaging

Drug application

Immunofluorescence labeling and confocal imaging

Western blots

Cellular viability assay

Mouse model of seizures

Acute seizures assessment and quantification

Statistical analysis

Footnotes

Acknowledgments

Author Disclosure Statement

Abbreviations Used

References

NMDAR-mediated neuronal Ca²⁺ influx

Effect of HMGB1 redox state on NMDA-induced Ca²⁺ increase

Single-cell Ca²⁺ imaging