Methamphetamine alters occludin expression via NADPH oxidase-induced oxidative insult and intact caveolae

Abstract Methamphetamine (METH) is a drug of abuse with neurotoxic and vascular effects that may be mediated by reactive oxygen species (ROS). However, potential sources of METH-induced generation of ROS are not fully understood. This study is focused on the role of NAD(P)H oxidase (NOX) in METH-induced dysfunction of brain endothelial cells. Treatment with METH induced a time-dependent increase in phosphorylation of NOX subunit p47, followed by its binding with gp91 and p22, and the formation of an active NOX complex. An increase in NOX activity was associated with elevated production of ROS, alterations of occludin levels and increased transendothelial migration of monocytes. Inhibition of NOX by NSC 23766 attenuated METH-induced ROS generation, changes in occludin protein levels and monocyte migration. Because an active NOX complex is localized to caveolae, we next evaluated the role of caveolae in METH-mediated toxicity to brain endothelial cells. Treatment with METH induced phosphorylation of ERK1/2 and caveolin-1 protein. Inhibition of ERK1/2 activity or caveolin-1 silencing protected against METH-induced alterations of occludin levels. These findings indicate an important role of NOX and functional caveolae in METH-induced oxidative stress in brain endothelial cells that contribute to the subsequent alterations of occludin levels and transendothelial migration of inflammatory cells.


Introduction
Methamphetamine (METH) is a highly addictive synthetic psychostimulant causing neurotoxicity in humans. METH neurotoxicity has been characterized by enhanced release and reduced synaptic reuptake of major monoamine neurotransmitters, including dopamine (DA) and serotonin , and is accompanied by a decrease in the number of dopamine transporter (DAT) binding sites. METH enters neurons via the DAT or 5-HT transporter and displaces both vesicular and intracellular DA and 5-HT. The displaced amines can then be oxidized producing reactive oxygen species (ROS). Indeed, increased tissue oxidative stress is known to be one of the major causes of neurotoxicity following METH administration [1,2]. However, cultured brain endothelial cells have also been shown to be a target for METH toxicity despite the lack of dopaminergic or serotonergic innervations. For example, exposure to METH can alter endothelial cell redox status by depleting cellular glutathione levels [3]. METH exposure leads to intracellular ROS generation in brain endothelial cells and disruption of blood-brain barrier (BBB) functions [4][5][6].
The BBB is a specialized system of capillary endothelial cells interconnected by intercellular tight junctions (TJs) that form a selectively barrier, which controls the internal environment of the central nervous system (CNS). TJs play a role in regulating the exchanges of substances between the brain and blood resulting in maintaining the homeostatic environment of the brain. TJs are constituted by transmembrane proteins such as occludin, claudins and junctional associated molecules (JAMs). In addition, cytoplasmic zonula occludens (ZO) proteins interact with transmembrane proteins and link them to the actin cytoskeleton [7].
Occludin, a 60-65 kD phosphoprotein, is highly expressed in cerebral endothelium, whereas it is much more sparsely distributed in peripheral endothelia [8]. It consists of four transmembrane domains, which bind to the two extracellular loops of claudin forming the paracellular component of the TJ. Mutation or overexpression of occludin in cultured cells affected both electrical resistance

NOX activity assay
Treated hCMEC/D3 cells were washed with ice-cold PBS and harvested in 500 l Tris-sucrose buffer (pH 7.1) containing 10 mM Tris, 340 mM sucrose, 1 mM EDTA and protease inhibitors mixture. Cell suspension was sonicated and centrifuged at 1475 ϫ g at 4°C for 15 min. to remove nuclei and unbroken cells. The supernatant was then centrifuged at 60,000 ϫ g at 4°C for 30 min. The pellet was suspended in 100 l of Tris-sucrose buffer, and NOX activity was evaluated by a lucigenin-enhanced chemiluminescence method as described by Sekhar et al. [21]. Intensity of chemiluminescence was measured using MikroWin2000 software and CentroXS3 LB 960 Microplate Luminometer (Berthold Technologies, Oak Ridge, TN, USA).

Caveolin-1 silencing
Caveolin-1 (Cav-1) silencing was performed as described previously [22] using small interfering RNA (siRNA) targeting human Cav-1 from Santa Cruz Biotechnology. Silencer Negative Control #1 siRNA (Applied Biosystems, Austin, TX, USA) was used as non-specific control siRNA. hCMEC/D3 cells were transfected overnight with 100 nM of Cav-1 or control siRNA using GeneSilencer siRNA Transfection Reagent (Genlantis, San Diego, CA, USA). Then, the cells were washed and allowed to recover for 2 days in normal medium before METH exposure.

Statistical analysis
Every experiment was independently performed at least three times. Data were expressed as the mean Ϯ S.E.M. One or two-way ANOVA was used to analyse the significance of the differences between the control and experimental groups. Value of P Ͻ 0.05 was considered significant.

METH alters occludin protein levels in cerebral endothelial cells
Disruption of BBB integrity is considered an important element of METH-induced neurotoxicity. Therefore, we first evaluated the effects of METH exposure (5- Time-dependent experiments (Fig. 1B) indicated that exposure to 10 M METH resulted in a relatively rapid decrease in occludin levels in NP40-soluble fraction. Indeed, decreased occludin levels were observed after 1 hr of METH exposure and remained reduced for the duration of treatment in this cellular fraction. Treatment with 10 M METH also altered occludin levels in NP-40 insoluble fraction; however, these changes were statistically significant only in cells exposed to METH for 6 hr.
To confirm and further characterize METH-induced changes, hCMEC/D3 cells were immunostained for occludin levels (Fig. 1C). Compared to control cells, cultures exposed to 10 M METH revealed a weaker occludin immunoreactivity that was markedly disrupted and discontinued at the areas corresponding to the cell-cell borders (arrows). We next determined the time-dependent  Figure 1D, diminished occludin levels were evident in membranes of cells exposed to METH between 1 and 6 hrs, reaching the lowest levels following a 3 hrs exposure and partially recovering 24 hrs post-METH treatment.
Another important transmembrane TJ protein is claudin-5 [7]. As determined by Western blotting, protein levels of claudin-5 were not affected by treatment with 10 M METH for up to 24 (data not shown).

Exposure to METH stimulates formation of an active NOX complex
Induction of oxidative stress is widely believed to be an important factor in METH-induced toxicity [1,2]. Because NOX is the major source of ROS in vasculature, we evaluated whether exposure to METH can stimulate NOX in hCMEC/D3 cells. Our initial studies concentrated on the phosphorylation of p47, which is essential for the activation of NOX. Western blot analysis indicated that p47 was rapidly phosphorylated upon METH treatment ( Fig. 2A). As determined by the densitometric analysis, the ratio of phosphorylated p47 (p-p47) to total p47 was significantly elevated from 30 min. to 3 hrs of METH exposure, that is, at the similar exposure time as METH-induced alterations of occludin levels.
Phosphorylation of cytoplasmic p47 on multiple serine residues induces its translocation to the membrane fraction and its interaction with membrane subunits of NOX, as well as phosphoinosistide lipid products [23]. To evaluate whether exposure to METH can result in the formation of an active NOX complex in hCMEC/D3 cells, p47 was immunoprecipitated with anti-p47 antibody followed by Western blot analysis of gp91 or p22. As indicated in Figure 2B, binding of p47 to gp91 and p22 was significantly increased by METH treatment. Enhanced immunoprecipitation of p47 with gp91 and p22 was induced in cells treated with METH for as short as 10 min. To ensure that the same protein amount was used for immunoprecipitation, GAPDH protein levels were measured as internal control in the supernatants obtained after precipitation of immune complexes.
To confirm the assembly of an active enzyme, we measured the NOX complex activity upon treatment with 10 M METH. As shown in Figure 2C, NOX activity was markedly increased after METH treatment with the maximal stimulation at 30 min., followed by the gradual return to the basal level at later time points. Additional treatment with NSC 23766, a specific inhibitor of NOX confirmed the specificity of the responses (Fig. 2D). Similar effects were also observed using other NOX inhibitors, diphenyleneiodonium chloride (DPI) and apocynin (data not shown).

NOX complex mediates METH-induced ROS generation
In the next series of experiments, we evaluated whether METHinduced activation of NOX can result in stimulation of oxidative stress in hCMEC/D3 cells using DHE fluorescence as the specific marker of superoxide production. DHE upon reaction with superoxide anions forms a red fluorescent product (ethidium), which intercalates with DNA [24,25]. In cells exposed to 10 M METH for 30 min., DHE fluorescence markedly increased (Fig. 3A, left). In FACS analysis, this effect corresponded to a shift of the histogram to the right (Fig. 3A, right, arrow). METH-induced cellular oxidative stress was confirmed using chloromethy-2', 7'-dichlorodihydrofluorescein diacetate (CM-H2DCFDA), a marker of a wide spectrum of ROS (data not shown).
To address the question of whether NOX activation is involved in METH-induced ROS generation, we pharmacologically inhibited NOX with NSC 23766 (100 M), followed by a 30 min. exposure to METH (10 M) and the assessment of DHE fluorescence by FACS analysis and confocal microscopy. As indicated in Figure 3B (left), inhibition of NOX significantly protected against the METH-induced DHE fluorescence. These effects were also confirmed under confocal microscopy as shown in Figure 3B (right), where a marked increase in red fluorescence corresponded to METH-induced formation of superoxide. Taken together, these observations provide evidence that ROS generated in hCMEC/D3 cells in response to METH treatment are mediated by an active NOX complex.

Inhibition of NOX protects against METH-induced alterations of occludin levels and transendothelial migration of monocytes
Our next series of experiments was devoted to investigating whether activation of NOX plays a role in METH-induced changes in occludin levels. hCMEC/D3 cells were pre-treated with 100 M NSC 23766, followed by 3 hrs exposure to 10 M METH. Figure 1, occludin levels were decreased by METH treatment in the NP40-soluble fraction with minimal changes in the NP-40-insoluble fraction. Cells exposed to METH in the presence of NSC 23766 exhibited significant protection against alterations of occludin levels in the NP-40 soluble fraction. However, exposure to NSC 23766 also resulted in redistribution of occludin as shown by decreased levels of this TJ protein in the NP-40 insoluble fraction of hCMEC/D3 cells (Fig. 4A).

Fig. 3 METH-induced superoxide production is mediated by an active NOX complex in hCMEC/D3 cells. (A) hCMEC/D3 cells were treated with 10 M METH for the indicated time period and the production of superoxide was measured by the oxidation of DHE as determined by flow cytometry. Relative superoxide levels (left) were expressed as percentage of control. (Right) Reflects an example of FACS analysis in hCMEC/D3 cells exposed to METH for 30 min. The arrow indicates a shift of histogram in the METH group indicating increased production of superoxide. (B) Inhibition of NOX protects against METH-induced production of superoxide. hCMEC/D3 cells were pre-treated with 100 M NSC 23766 (NSC), followed by exposure to 10 M METH for 30 min. Production of superoxide (left) was quantified by DHE staining and FACS analysis as in (A) (left) and visualized under the confocal microscope (right). Scale bar, 50 m. The values in (A) and (B) represent the mean Ϯ S.E.M., n ϭ 3; *Statistically significant as compared to the respective control. † Values in the METHϩNSC group are significantly different as compared to those in the METH group.
We also determined the effects of NSC 23766 on METHinduced alterations of protein levels of occludin using immunofluorescence labelling (Fig. 4B, left). Analysing the obtained data, we specifically focused on occludin immunoreactivity in the areas corresponding to the cell-cell borders, which was then quantified and plotted in a form of a bar graph (Fig. 4B, right). Exposure to NSC 23766 alone did not affect occludin immunoreactivity. Notably, METH treatment significantly reduced occludin-positive staining; however, these effects were fully reversed by NSC 23766 pre-treatment.

Integrity of occludin may regulate the barrier functions of brain endothelial cells and paracellular migration of inflammatory cells. Therefore, we also determined the involvement of NOX in METHinduced monocyte migration across monolayers of brain endothelial cells. Treatment of hCMEC/D3 cells with 10 M of METH for 5 hrs resulted in a 33% increase in transendothelial monocyte
passage. Pre-treatment with NSC 23766 protected against these effects (Fig. 4C), demonstrating the role of NOX in METH-induced barrier dysfunction of brain endothelial cells. In addition, treatment with NSC 23766 alone significantly decreased transendothelial migration of monocytes.

Functional caveolae and ERK1/2 signalling are detrimental for METH-induced ROS production and alterations of occludin levels
An active NOX complex was shown to be associated with caveolae [26]. In addition, signalling pathways localized to caveolae may be involved in the regulation of TJ proteins [22]. Therefore, in the last series of experiments, we determined the involvement of functional caveolae in METH-induced changes in signalling pathway activation and occludin protein levels. Treatment with METH induced a rapid (within 10 min.) phosphorylation of ERK1/2 without changing the total level of this kinase. In addition, a 30 min. exposure to METH increased phosphorylation of Cav-1 protein, the structural and regulatory protein in caveolae. The antibody used in these experiments specifically recognized phosphorylation at Tyr14. Interestingly, inhibition of ERK1/2 activity by a specific inhibitor U0126 completely protected against METH-induced phosphorylation of Cav-1, indicating that this process is dependent on active ERK1/2 (Fig. 5A).
To address the role of caveolae in these events, hCMEC/D3 cells were transfected with Cav-1 siRNA. This procedure resulted in a decrease of Cav-1 protein to negligible levels and prevented phosphorylation of Cav-1 (Fig. 5B). In contrast, Cav-1 silencing did not affect ERK1/2 phosphorylation, indicating that ERK1/2 activation is upstream from Cav-1 phosphorylation. The role of ERK1/2 signalling in METH-induced alterations of TJ proteins was confirmed by the observation that pre-treatment with U0126 protected against altered occludin levels (Fig. 5C). Interestingly, Cav-1 silencing also prevented METH-induced changes in occludin levels (Fig. 5D), indicating the role of functional caveolae in the regulation of TJ protein levels.

Discussion
METH is a frequently abused drug via the intranasal, intravenous, smoked and, less commonly, the oral route [27]. Although the onset of psychostimulatory effects of METH depends on the route of exposure, METH plasma concentrations continuously increase for 4 hrs post a single drug administration with the mean plasma METH values ~0.3 M after administration of 15 mg METH [28]. Higher mean plasma METH values were reported following METH vapour inhalation, reflecting faster absorption via smoking [29][30][31]. However, METH is commonly abused in multiple dose cycles, with an interdose interval of 0.5-3 hrs, which may continue for several days [31,32]. Indeed, METH abusers typically use 20-40 mg METH more than once a day for approximately 15-20 days per month [33]. METH body burden in abusers is estimated at ~50 mg, with the blood concentrations in the range of 0. 1-11.1 M [34].
Although METH toxicity has recently been reported using different cell systems, several of these studies employed very high concentrations of METH that markedly exceeded the plasma levels of METH abusers. For example, treatment with 1.68 mM METH induced intracellular oxidative stress and mitochondrial alterations in a human dopaminergic neuroblastoma SH-SY5Y cell line [35] and 4 mM METH induced similar effects in cultured mouse astrocytes [36]. Similarly, generation of ROS and toxicity to human brain endothelial cells were demonstrated using METH at the dose of 2.5 mM [6]. Thus, this study is unique in terms of using an METH concentration of 10 M that is relevant to human drug

abuse. Using sensitive detection methods, we demonstrated that such pathologically relevant doses of METH can induce intracellular oxidative stress and alter TJ integrity in human cerebral microvascular endothelial cells.
A wildly recognized mechanism of METH-induced pathology is related to stimulation of cellular oxidative stress and several studies implicated ROS as an important mediator of BBB disruption and neurotoxicity induced by this drug. The intracellular sources of ROS generated by METH are not fully understood; however, mitochondrial toxicity has been considered [37]. Indeed, exposure to METH decreased mitochondrial membrane potential, increased mitochondrial mass, enhanced protein nitrosylation and diminished protein levels of complexes I, III and IV of the electron transport chain in primary human T cells. These changes were associated with impaired functions of T cells. Importantly, antioxidants attenuated METH-induced mitochondrial damage, further indicating a cross-relationship between METH cytotoxicity, mitochondrial damage and cellular oxidative stress [37]. Another potential source of METH-induced ROS is activation of cytochrome P-450 (CYP450) [38].
Novel results of this study focus on the role of METH-induced activation of NOX, a ROS-producing enzyme that was originally identified and characterized in phagocytes and is also expressed within the cerebral vasculature [39]. Similar to the leukocyte form, NOX expressed in endothelial cells contains both gp91 and p22 [19], that is, the subunits that binding to p47 was increased in hCMEC/D3 cells in response to METH treatment in this study (Fig. 2). Excessive generation of superoxide and its derivatives by NOX within the brain microvasculature may cause lipid peroxidation and membrane disruption, thus damaging BBB integrity [40]. Indeed, our studies revealed that a 30 min. treatment with 10 M METH can stimulate the formation of NOX complex, associated with increased enzyme activity and generation of ROS in hCMEC/D3 cells. To prove the direct involvement of NOX in METH-induced ROS generation, we inhibited this enzyme activity with NSC 23766. As indicated in Figure [41], our immunostaining results do not support this notion. In fact, occludin-positive immunoreactivity was markedly diminished and disrupted in the areas corresponding to the cell-cell borders (Figs 1C and 4B). Moreover, occludin protein levels were significantly decreased in the membranes extracted by a commercially available kit (Fig. 1D).
Important results of this study provide evidence that inhibition of NOX by NSC 23766 can protect against METH-induced diminished occludin levels. These results are in agreement with the reports showing the role of cellular oxidative stress in alterations of TJ expression [42,43]. It was demonstrated that increased oxidative stress activated protein tyrosine kinases [44], RhoA, and PI3 kinase [45], resulting in the disruption of the barrier properties of brain endothelial cells and enhanced permeability across the in vitro model of the BBB. Part of this sequence of events has been linked to the activation of myosin light chain kinase, indicating the role of cytoskeleton in METH-induced vascular toxicity [5]. It should be pointed out that NSC 23766 blocks NOX by inhibition of Rac1, leading to activation of the Rho pathway [46]. Although the Rho signalling is linked to TJ assembly [47], alterations of Rho activity may contribute to a decrease in monocyte migration as observed in hCMEC/D3 cells cocultured with monocytes and exposed to this drug (Fig. 4C). Modulation of the Rho pathway may also be responsible for occludin redistribution in the NP-40 insoluble fraction of hCMEC/D3 cells (Fig. 4A).
Previous results from our laboratory emphasized the role of degradative processes, namely activation of matrix metalloproteinases and proteasome, in degradation of TJ proteins [48]. In addition, an interesting mechanism of caveolae-mediated internalization of occludin was recently described as a process associated with CCL2-induced TJ remodelling in brain endothelial cells [49]. It was observed that within a 1 hr CCL2 treatment, TJ proteins containing occludin internalized via a caveolae-dependent pathway into recycling endosomes. Internalized TJ proteins did not appear to undergo degradation and TJ recycling contributed to TJ complex recovery. This alternative process may also contribute to METH-induced alterations of occludin levels in the cellular membrane fraction observed in this study.
Another possibility of METH-induced alterations of TJ complexes might be related to induction of inflammatory responses and increased levels of pro-inflammatory cytokines, which can activate the ERK1/2 pathway and thus influence phosphorylation of TJ proteins. Indeed, exposure to METH results in increased levels of IL-1␤, IL-6 and TNF␣, as shown both in whole brain homogenates and in individual cell types [2,3,50,51]. However, an increase in cytokine levels is typically observed a few hours post-METH exposure. In contrast, we observed the maximum of ERK1/2 phosphorylation already 10 min. following METH treatment. This time-dependent sequence suggests that activation of the MAPK pathway might be the effect of METH via NOX activation rather than a secondary effect mediated by cytokine induction.
The ultimate outcome of the disrupted integrity of the BBB can be enhanced endothelial permeability and elevated transendothelial leukocyte migration causing subsequent tissue damage. Therefore, we also evaluated the effects of METH on transendothelial migration of U937 cells. Consistent with alterations of occludin protein levels, exposure to METH induced an increase in migration of human monocytic cells across hCMEC/D3 monolayers. These effects appeared to be mediated by METH-induced activation of NOX, because an inhibition of this enzyme protected against enhanced cell migration. Overall, these results demonstrate a functional significance of 'leaky' barrier properties of brain endothelial cells because of altered occludin levels that may stimulate paracellular transendothelial migration of leukocyte and thus contribute to METH-induced neuroinflammatory responses. Nevertheless, exposure to METH is known to induce a variety of redox-responsive transcription factors and pro-inflammatory genes that also can influence adhesion and transendothelial migration of inflammatory cells.
Caveolae are a subgroup of lipid rafts abundant in endothelial cells that play a role in the regulation of various endothelial functions [52]. For example, caveolae and caveolae-associated pathways have emerged as regulators of TJ integrity [22] and the platform necessary for the assembly and activation of NOX [26]. Cav-1 protein was initially identified as a tyrosine-phosphorylated substrate of v-src [53]; however, recent evidence indicates that it can play a regulatory role in caveolae-associated signalling and organize the association of signalling molecules within caveolae [54]. Down-regulation of Cav-1 expression either by gene silencing or knock-out approaches eliminates functional caveolae in endothelial cells.
Novel results of this study demonstrate that treatment with METH stimulates phosphorylation of Cav-1 in tyrosine 14 residue (Tyr14). Although literature data suggest that Cav-1 Tyr14 is a principal target for Src kinase phosphorylation [53], this study clearly indicates that Cav-1 phosphorylation at Tyr14 is regulated by the ERK1/2 pathway in response to METH exposure. Indeed, ERK1/2 inhibition completely protected against METH-induced phosphorylation of Cav-1 at Tyr14 (Fig. 5A). These results are important because phosphorylation of Cav-1 appears to be involved in the regulation of paracellular permeability [55]. For example, an increase in Cav-1 phosphorylation was determined in endothelial cells exposed to hydrogen peroxide and protection against this effect attenuated hydrogen peroxide-induced hyperpermeability of endothelial cell monolayers [56].
This study also indicates the role of functional caveolae in the regulation of METH-induced alterations of occludin protein levels. Specifically, Cav-1 silencing completely protected against decreased occludin levels in METH-treated hCMEC/D3 cells (Fig. 5D). Similar protections against alterations of TJ protein levels were observed in brain endothelial cells isolated from Cav-1 deficient mice and exposed to HIV-1 protein Tat [22]. Interestingly, blocking ERK1/2 activation by U0126 also protected against METH-induced alterations of occludin levels (Fig. 5C).
In conclusion, this study demonstrates that METH, at the levels relevant to human abuse, can activate a NOX complex, followed by the subsequent activation of ERK1/2 signalling and phosphoryla-tion of Cav-1 at Tyr14. Next, these events result in the generation of ROS, alterations of occludin protein levels, and compromised barrier function of brain endothelial cells (Fig. 6). Disruption of functional caveolae and/or inhibition of Cav-1 phosphorylation protected against METH-induced production of ROS and disruption of occludin levels. These results suggest that targeting NOX and caveolae-associated pathways may provide an important therapeutic strategy against METH-induced vascular toxicity. Fig. 6 Schematic diagram of METH-induced alteration of occludin protein levels and increased monocyte transendothelial migration. METH in concentrations relevant to drug abuse in humans activates NOX by a caveolaeassociated mechanism. Activated NOX results in increased ROS generation with the subsequent phosphorylation of ERK1/2 and Cav-1, followed by alterations of occludin levels. An increase in transendothelial migration of monocytes appears to be an ultimate outcome of these events.