Diclofenac Enhances Proinflammatory Cytokine-Induced Aquaporin-4 Expression in Cultured Astrocyte

Abstract

Acute encephalopathy is a generic term for acute brain dysfunction occurring after infection. Acute encephalopathy induced by influenza virus results in high mortality, and most cases of influenza-associated encephalopathy (IAE) result in brain edema. Administration of diclofenac sodium (DCF), a non-steroidal anti-inflammatory drug (NSAID), is associated with a significant increased mortality rate of IAE. These previous clinical findings proposed further investigation of DCF administration and brain edema to clarify how DCF aggravates IAE. Aquaporin-4 (AQP4) is the predominant water channel protein in the mammalian brain, and is mainly expressed in astrocytes. AQP4 plays an important role in brain water homeostasis. Therefore, we investigated a possible association between DCF and AQP4 production in astrocytes. We stimulated cultured rat astrocytes with three cytokines, interleukin-1β, tumor necrosis factor α, and interferon γ, and then treated with DCF. DCF enhanced proinflammatory cytokine-induced AQP4 gene and protein expression in astrocytes, whereas DCF alone did not change the AQP4 gene expression. The addition of nuclear factor-kappa B (NF-κB) inhibitor abrogated AQP4 gene and protein expression completely in astrocytes treated with cytokines alone and in those also treated with DCF. In conclusion, this study demonstrated that AQP4 is upregulated in astrocyte by proinflammatory cytokines, and that the addition of DCF further augments AQP4 production. This effect is mediated via NF-κB signaling. The enhancement of AQP4 production by DCF may explain the significantly increased mortality rates in IAE patients treated with DCF.

Introduction

Infection with influenza virus has been associated with a wide variety of neurological complication, including seizures, Reye’s syndrome, encephalopathy/encephalitis, acute necrotizing encephalitis, acute hemorrhagic encephalopathy, and transverse myelitis (Maricich et al. 2004). Influenza-associated encephalopathy (IAE) is one of the most severe neurological diseases in children, particularly those under 5-years-old. During the acute stage of IAE, influenza virus is rarely detected in the cerebrospinal fluid (CSF); however, several immune mediators and cytokines, including nitric oxide (NO), tumor necrosis factor α (TNFα), interleukin-1β (IL-1β), and interferon gamma (IFNγ) are elevated in the serum and CSF of patients with IAE (Kawashima et al. 2003, 2005; Kawada et al. 2006; Yamanaka et al. 2006; Mizuguchi et al. 2007). Many researchers therefore attribute the severe pathogenesis of IAE to hyperimmunization and a central nervous system (CNS) cytokine storm (Mizuguchi et al. 2007). The fact that, and most cases are complicated by brain edema (Mizuguchi et al. 2007).

Drugs such as aspirin, a non-steroidal anti-inflammatory drug (NSAID), and valproic acid have also been associated with encephalopathy (Powell-Jackson et al. 1984; Belay et al. 1999; Schrör 2007), and in Japan, use of the NSAID, diclofenac sodium (DCF), is associated with a significant increase in the mortality rate of patients with IAE (Mizuguchi et al. 2007). Consequently, DCF usage is prohibited for use, with acetaminophen (N-acetyl-p-aminophenol, APAP) recommended as an antipyretic and an analgesic for children with influenza virus infection (Mizuguchi et al. 2007; Nagao et al. 2008). The mechanism by which DCF aggravates IAE remains unclear and investigations into the function of DCF could have important implications for clarifying the pathogenesis of IAE.

Brain edema is associated with many intracranial neuropathological states and metabolic diseases (Gunnarson et al. 2004; Kimelberg 2004; Unterberg et al. 2004). Although a number of studies have addressed the underlying molecular mechanisms of brain edema, it is still unclear how regulation of water transport takes place across the blood–brain barrier (BBB), the CSF–brain interface, and between extracellular and intracellular compartments in brain parenchyma (Lee et al. 2008). Aquaporin-4 (AQP4) is a predominant water channel protein in the mammalian brain, and is expressed in the endfeet of astrocytes surrounding brain capillaries and comprising the BBB (Amiry-Moghaddam and Ottersen 2003b; Yukutake and Yasui 2010). AQP4 has also been associated with the development of acute brain edema (Manley et al. 2000; Vajda et al. 2002; Bloch et al. 2005).

In this study, we investigated whether DCF enhances AQP4 expression in cultured primary rat astrocytes stimulated by proinflammatory cytokines, IL-1β, TNFα, and IFNγ. Examining whether DCF affects AQP4 expression in astrocytes under inflammatory conditions could further clarify the pathogenesis of IAE.

Materials and Methods

Astrocyte Culture

Astrocytes were prepared from the cerebral cortex of postnatal day 1 or 2 Wistar rats as described (Miyachi et al. 2001). The cerebral cortex was dissected, trypsinized, and then dissociated in low-glucose (1 g/l) Dulbecco’s modified Eagle’s medium (l-DMEM; Invitrogen, Carlsbad, CA) supplemented with 10 % fetal bovine serum (FBS) (Invitrogen), penicillin (80 units/ml), and streptomycin (0.2 mg/ml). Dissociated cells were placed into 25-cm² culture flasks (Corning Incorporated Life Sciences, Lowell, MA), and after 7 to 10 days, cultured glial components were dislodged with 0.1 % trypsin and replated onto 60-mm FALCON tissue culture dishes (1–1.5 × 10⁶ cells/dish) or 24-well FALCON plates (0.5–1.0 × 10⁵ cells/well). After secondary culture for 5 to 7 days, the medium was replaced with high-glucose (4.5 g/l) DMEM (h-DMEM) containing 1 % N-2 supplement (Invitrogen) for experiments. Immunocytochemistry with anti-glial fibrillary acidic protein (GFAP) antibodies (DAKO, Glostrup, Denmark) revealed that >95 % of the cell population was astrocytes.

Cell Stimulation

We used three proinflammatory cytokines, IL-1β, TNFα, and IFNγ, as described (Kozuka et al. 2005, 2007). After replacing the culture medium with fresh medium, cells were stimulated with rat IL-1β (5 ng/ml), rat TNFα (20 ng/ml), and rat IFNγ (5 ng/ml) (Wako Pure Chemical Industries, Osaka, Japan). When used, DCF (1 μg/ml; Wako), indomethacin (IND; 1 μg/ml; Sigma), or ibuprofen (IB; 1 μg/ml; Sigma) was added to the culture medium as COX inhibitors. We determined the drug concentration of each COX inhibitor in the present experiments according to the maximum therapeutic concentration of each COX inhibitor in serum. Cells were maintained in a humidified chamber at 21 % O₂.

Quantitative Reverse Transcription-Polymerase Chain Reaction (Q-RT-PCR)

Q-RT-PCR was performed on selected genes using the 7700 Sequence Detection System (Applied Biosystems, Foster City, CA). Total RNA was isolated using TRIzol™ reagent (Invitrogen) according to the manufacturer’s protocol. cDNA was obtained from total RNA samples. Reverse transcription (RT) was carried out using random primers and Ready-To-Go You-Prime First Strand Beads (GE Healthcare Bio-Science Corp., Piscataway, NJ), and the resulting cDNA was subjected to PCR-based amplification. Primer pairs used for amplification were as follows:

• AQP4: forward, 5′-TTATACCGGAGCCAGCATGAATC-3′,

• AQP4: reverse, 5′-AAGTGCACCTGCCAGCACA-3′;

• β-actin: forward, 5′-TCATGAAGTGTGACGTTGACATCCGT-3′,

• β-actin: reverse, 5′-CCTAGAAGCATTTGCGGTGCAGGATG-3′.

Real-time PCR was performed using SYBR Green PCR Master Mix Reagents (Applied Biosystems). Amplification was performed by activation of AmpliTaq Gold DNA polymerase at 95 °C for 10 min, followed by amplification for 40 cycles at 95 °C for 15 s and 60 °C for 1 min. The amount of RNA was calculated with relative standard curves for each mRNA of interest and β-actin. Normalization to β-actin was conducted to account for variability in the quality and total concentration of RNA, as well as for RT efficiency.

Western Blotting

After treatment, cells were harvested and gently homogenized on ice using 5 mM Tris–HCl (pH 7.4) containing 4 M urea, 0.5 % Nonidet P-40 (NP-40), and Protease Inhibitor Cocktail (Sigma, St Louis, MO). Protein content was estimated by the bicinchoninic acid method described by Pierce Chemical (Rockford, IL). The protein samples were suspended in sodium dodecyl sulfate (SDS) sample buffer. Equal amounts of proteins were separated under denaturing conditions by electrophoresis in a 7.5 % polyacrylamide gel containing SDS and electrotransferred onto polyvinylidene difluoride (PVDF) membrane (Immobilion-P, Millipore, Billercia, MA). The membrane filters were blocked with 5 % skim milk in TBS-T (20 mM Tris–HCl (pH 7.6), 137 mM NaCl, 0.1 % Tween 20) at 4 °C overnight, followed by incubation with polyclonal anti-AQP4 (1:500, Millipore) diluted in TBS-T at 4 °C overnight. The blots were developed using the appropriate secondary antibodies conjugated to horseradish peroxidase (1:5,000, Amersham Biosciences Corp., Piscataway, NJ), and bands were visualized using an enhanced chemiluminescence method (ECL, Amersham Biosciences Corp.). For normalization of protein loading, blots were stripped and reprobed with polyclonal anti-actin (1:1,000, Sigma) in TBS-T. Relative band intensities were determined by densitometry using Kodak Digital Science 1D version 2.0 (Eastman Kodak Company, Rochester, NY).

Immunocytochemical Staining

Cells grown on glass coverslips coated with BD Matrigel Matrix (BD Bioscience, San Jose, CA) were washed with phosphate-buffered saline (PBS), fixed in 3 % paraformaldehyde at room temperature for 30 min, permeabilized with 0.2 % Triton X-100 for 5 min, washed with PBS, and blocked for 1 h at room temperature with blocking solution (3 % bovine serum albumin, 0.1 % glycine in PBS). After washing, cells were labeled with primary antibodies against AQP4 (1:500, Millipore) and against glial fibrillary acidic protein (GFAP; 1:500, Millipore) for 1 h, then with fluorescence-conjugated secondary antibodies for 1 h. Alexa Fluor 488-labeled (green) goat anti-rabbit IgG and Alexa Fluor 594-labeled (red) goat anti-mouse IgG (Invitrogen) were used as the secondary antibodies. After washing, the cells were mounted on glass slides with ProLong Gold antifade reagent and stained with 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen). Stained cells were examined using an AX70 fluorescence microscope (Olympus, Tokyo, Japan). For quantification, the number of AQP4-positive cells per field was counted at 400× magnification randomly. Only cells showing precise cytoplasmic staining of AQP4 antibody were counted as positive cells to avoid counting non-specific cell staining. The counts from six fields were averaged.

Statistical Analysis

Statistical analyses were performed using SPSS for Windows version 13.0 (SPSS, Chicago, IL). Analysis of variance (ANOVA) was used to compare continuous data, followed by Bonferroni’s post hoc test. Data are reported as mean ± SEM. P < 0.05 was considered statistically significant.

Results

Expression of AQP4 in Cultured Astrocytes Stimulated by Proinflammatory Cytokines and DCF

We investigated the effects of COX inhibitors (DCF, IB, and IND) on the induction of AQP4 mRNA expression when used in combination with cytokine stimulation. Only DCF in combination with cytokine stimulation increased AQP4 gene expression to a level threefold greater than that seen with cytokine stimulation alone (P < 0.05). IB slightly enhanced AQP4 gene expression following cytokine stimulation, while IND had no effect on AQP4 mRNA expression (Fig. 1). Thus, of the various COX inhibitors, only DCF seems to induce AQP4 in cultured astrocytes when combined with proinflammatory stimulation.

Fig. 1.

Effects of the COX inhibitors DCF, indomethacin (IND), and ibuprofen (IB) on AQP4 gene expression in cultured astrocytes stimulated with three proinflammatory cytokines (Cyto: IL-1β, TNFα, and IFNγ). Effects of each of the COX inhibitors on AQP4 gene expression, as determined by quantitative reverse transcription-polymerase chain reaction (Q-RT-PCR). Cells were exposed to each of the drugs for 12 h. AQP4 mRNA expression is normalized against that of β-actin. Bars indicate AQP4 expression relative to that in control astrocytes. Data represents mean ± SEM.; n = 6. ^‡ P < 0.05, compared to control. *P < 0.05, compared to each drug alone. ^# P < 0.05, compared to cytokines alone

Subsequent time-course analysis of the AQP4 gene expression revealed that cytokine-stimulated astrocytes expressed significantly more AQP4 than control astrocytes, but that no significant difference in expression was apparent between control astrocytes and those treated with DCF alone across the time scale. However, DCF applied in combination with cytokines increased the gene expression of AQP4 by fivefold compared with control or DCF alone, and by threefold compared with the cytokine-only treated cells at 12 h (Fig. 2). AQP4 gene expressions also peaked later in astrocytes stimulated with DCF and cytokines than in those treated with cytokines alone (Fig. 2). Western blotting showed that AQP4 protein levels at 24 h were approximately 1.8-fold greater in astrocytes following treatment with DCF and cytokines, and approximately 1.5-fold greater in astrocytes following treatment with cytokine alone, compared to control astrocytes and astrocytes treated only with DCF (Fig. 3a, b). These data suggested that DCF works synergistically with cytokines to enhance the AQP4 expression in astrocytes.

Fig. 2.

Effect of DCF on AQP4 gene expression in cultured astrocytes stimulated with three proinflammatory cytokines (IL-1β, TNFα, and IFNγ) and/or DCF. Changes in AQP4 gene expression in cultured astrocytes stimulated with three cytokines and DCF (1 μg/ml) were measured by Q-RT-PCR. Expression levels were normalized to β-actin gene expression. Bars indicate AQP4 expression relative to that in control astrocytes. Data represents mean ± SEM; n = 7. ^‡ P < 0.05, compared to control. *P < 0.05, compared to DCF alone. ^# P < 0.05, compared to cytokines alone

Fig. 3.

Western blot analysis of AQP4 protein in cultured astrocytes stimulated for 12 h (a) and 24 h (b) with DCF, three cytokines, or DCF with cytokines. An AQP4-specific antibody was used to measure AQP4 levels and then the blot was reprobed for actin to ensure equal loading. Semi-quantitative measurements of AQP4 protein levels were performed using NIH Image software. AQP4 expression was normalized to actin expression. Bars indicate AQP4 expression relative to that in control astrocytes. Values are expressed as mean ± SEM (n = 5). ^‡ P < 0.05, compared to control. *P < 0.05, compared to DCF alone. ^# P < 0.05, compared to cytokines alone

Expression of the AQP4 in Cultured Astrocytes Treated with Cytokines and DCF is Mediated Via Nuclear Factor-Kappa B (NF-κB)

Previously we reported that inducible nitric oxide synthase (iNOS) gene expression in cultured astrocytes treated with cytokines and DCF is mediated via NF-κB, and not via the cAMP/PKA cascade (Kakita et al. 2009). To examine whether NF-κB signaling also mediates AQP4 upregulation, we used BAY11-7082 (10 μM, Millipore) to inhibit NF-κB signaling. As shown in Fig. 4a, this NF-κB inhibitor suppressed AQP4 gene expression in cytokine-treated astrocytes, with or without DCF, to the expression levels in control astrocytes. Western blot analysis showed that AQP4 protein levels were also suppressed significantly to the expression level in control astrocytes (Fig. 4b, c). These data suggested that DCF induces the gene and protein expression of AQP4 via NF-κB signaling.

Fig. 4.

Effects of NF-κB inhibitors on AQP4 induction in cytokine-stimulated cultured astrocytes. a Astrocytes were treated with NF-κB inhibitors, BAY11-7082 10 μM 30 min before treatments and then treated with cytokines and/or DCF for 12 h and AQP4 expression was measured by Q-RT-PCR. Expression levels were normalized to that of β-actin. Relative AQP4 expression indicates AQP4 expression relative to that in control astrocytes (mean ± SEM; n = 7). b Astrocytes were treated with NF-κB inhibitors, BAY11-7082 10 μM 30 min before treatments and then treated with cytokines and/or DCF for 24 h. AQP4 protein expression was measured by western blotting. An AQP4-specific antibody was used to measure AQP4 levels and then the blot was reprobed for actin to ensure equal loading. c Semi-quantitative measurements of AQP4 protein levels were performed using NIH Image software. AQP4 expression is normalized to actin expression. AQP4 expression is relative to that in control astrocytes (mean ± SEM; n = 5). ^‡ P < 0.05 compared to astrocytes stimulated with cytokines without the inhibitor. *P < 0.05 compared to astrocytes stimulated with DCF and cytokines without the inhibitor

Immunocytochemical Analysis of AQP4 Expression in Cultured Astrocytes

AQP4 and GFAP expressions were examined in cultured cortical astrocytes by immunocytochemistry. AQP4 protein was localized to cytoplasm diffusely and was not apparent on the cell membranes (Fig. 5a, d, g, j). In control cells and cells treated with DCF, AQP4 staining was observed in the perinuclear region and diffusely throughout the cytoplasm (Fig. 5a, d). Treatment with cytokines, with and without DCF, increased the AQP4 protein in astrocytes with higher intensity staining visible in most cells (Fig. 5g, j). GFAP was localized to fibers in the cytoplasm (Fig. 5b, e, h, k). There was no significant difference in GFAP staining across the four cell groups examined (Fig. 5b, e, h, k). To confirm this finding, we counted the number of AQP4-positive cells with precise cytoplasmic staining and confirmed that cytokine stimulation significantly increased the number of AQP4-positive cells compared with control astrocytes, and that AQP4-positive cell numbers were not significantly different between astrocytes treated with DCF alone and control astrocytes. DCF addition seemed to enhance the number of AQP4-positive cells in cytokine-stimulated astrocytes (Fig. 5m).

Fig. 5.

Immunocytochemical detection of AQP4 in individual astrocyte stimulated by the three cytokines, DCF, or DCF with cytokines for 12 h. Staining with anti-AQP4 antibody (green) (a, d, g, j), cell marker anti-GFAP antibody (red) (b, e, h, k), and merged image (c, f, i, l). Nuclei were stained with DAPI (blue). Scale bar represents 50 μm. m Only cells stained precisely in cytoplasm with AQP4 antibody were counted as AQP4-positive cells to avoid counting non-specific staining. DAPI-stained cells were used to estimate total cells. The counts from six fields were averaged. Bar indicate the ratio of AQP4 positive cells per DAPI positive cells. Data represents mean ± SEM; n = 6. ^‡ P < 0.05, compared to control. *P < 0.05, compared to DCF alone. ^# P < 0.05, compared to cytokines alone

Discussion

Several recent studies have drawn attention to the importance of a well-controlled brain water homeostasis for the stability of neuronal function (Binder and Steinhauser 2006; Seifert et al. 2006). Perturbations in this homeostasis are common in acute brain diseases including stroke, trauma, and meningitis, and are a major cause of permanent brain damage (Saez-Llorens and McCracken 2003; Unterberg et al. 2004). It is well known that astrocytes play an important role in brain water homeostasis, as the predominant expression sites of the major brain water channel protein AQP4. A previous study demonstrated a novel astrocyte-specific enhancer, involving the POU transcription factors, in the 5′-flanking region upstream of exon 0 of the AQP4 gene, and showed that this transcriptional regulation was activated only in astrocytes (Abe et al. 2012).

Brain edema can be classified into cytotoxic and vasogenic types (Papadopoulos and Verkman 2007). Cytotoxic edema occurs when fluid flows accumulate from the vascular compartment into the intracellular space through an intact BBB, while in vasogenic edema, the BBB becomes leaky, permitting the entry of plasma fluid into the brain. AQP4 is modulated with acute cytotoxic brain edema after acute brain injury and AQP4 null mice are protected from several models of cytotoxic brain edema including hyponatremia (Papadopoulos and Verkman 2007), bacterial meningitis (Papadopoulos and Verkman 2005), and focal cerebral ischemia (Manley et al. 2000). On the other hand, the AQP4 null mice are more sensitive to vasogenic edema due to the increased permeability of capillary endothelial cells associated with brain tumors (Papadopoulos et al. 2004). In the tumor model, the AQP4 null mice developed a worse neurological score and higher intracranial pressure than wild-type controls. These data therefore suggest that AQP4 expression or function could be beneficial or harmful in the presence of brain edema, depending on the type. Recently the usage of NSAIDs, especially DCF, was associated with a significant increase in the mortality rate of IAE in Japan (Mizuguchi et al. 2007; Nagao et al. 2008). These clinical studies proposed that further investigation linking the usage of DCF and brain edema might provide important insights into how DCF aggravates IAE. This study demonstrated that DCF enhances the expression of AQP4 mRNA and protein in cultured astrocytes following clinically relevant cytokine treatment (Kawashima et al. 2003, 2005; Kawada et al. 2006; Yamanaka et al. 2006; Mizuguchi et al. 2007). Interestingly, DCF used in the absence of cytokine stimulation did not increase the AQP4 mRNA and protein expression in cultured astrocytes (Figs. 2, 3), providing further evidence that inflammatory mediators could play important roles in the induction of AQP4 expression by DCF in astrocytes.

DCF is a non-selective cyclooxygenase (COX) inhibitor that inhibits prostaglandin E2 (PGE2) synthesis (Hinz and Brune 2002). COX, also known as prostaglandin H synthase, catalyses the first committed step in the synthesis of prostanoids, a large family of arachidonic acid metabolites comprising prostaglandins, prostacyclin, and thromboxanes (Minghetti 2004). The non-selective COX inhibitors inhibit the production of PGE2 and are widely used as antipyretics and analgesics (Hinz and Brune 2004). We previously reported that DCF acts synergistically with the elevated proinflammatory cytokines in the serum and CSF of patients with IAE, by producing increased iNOS and NO in astrocytes, which in turn leads to cell damage (Kakita et al. 2009). In this previous study, proinflammatory cytokines induced COX2 expression, but not COX1 expression, in cultured astrocytes with and without DCF. The addition of DCF did not enhance COX2 expression and synthesis of PGE2 in cultured astrocytes stimulated by proinflammatory cytokines. Otherwise, DCF might upregulate iNOS gene expression due to increased duration of NF-κB activation during inflammation. These data suggested that proinflammatory cytokine stimulation might alter the pharmacological effect of DCF in astrocytes (Kakita et al. 2009). Thus, we examined whether the increased AQP4 expression was induced via NF-κB signaling. The NF-κB inhibitor, BAY11-7082, inhibited AQP4 gene and protein expression significantly in astrocytes stimulated by cytokines alone and in those treated with cytokines and DCF (Fig. 4). Our previous study demonstrated that the phosphorylation of NF-κB p65 and the degradation of IκB allowed NF-κB to translocate into the nucleus in cultured astrocytes in the presence of cytokine stimulation, and that this effect was enhanced by the addition of DCF (Kakita et al. 2009). These results suggested that DCF induces AQP4 expression in cytokine-stimulated astrocytes via NF-κB signaling, similar to iNOS/NOx induction in astrocytes.

The present immunocytochemical analysis showed that AQP4 protein was localized diffusely to cytoplasm and was not apparent on cell membranes. However, a previous immunohistochemical study localized AQP4 around vessels in the brain cortex and to the foot process of astrocytes surrounding vessels, using the same antibody, in our previous study (Aoyama et al. 2012). These findings could indicate that AQP4 protein localization differs between astrocytes in the brain and cultured astrocytes. As many researchers know, astrocytes lose its structural polarity and change to flat and muriform shape in culture. Moreover, AQP4 is restrictively expressed in astrocyte endfeet contacting the blood vessel, not throughout the membrane (Nielsen et al. 1997; Amiry-Moghaddam et al. 2003a; Papadopoulos and Verkman 2007; Lee et al. 2008). These results indicate that some mechanism to control AQP4 localization exists in structurally polar astrocytes, and it might critically associate with water permeability in BBB. Further study is needed to clarify the change of AQP4 protein localization in astrocytes under the pathological condition.

In conclusion, we revealed that DCF enhances proinflammatory cytokine-induced AQP4 gene and protein expression in cultured astrocytes. These results might help to explain why the usage of DCF during IAE leads to the exacerbation of brain edema and a significant increase in the mortality rate of IAE.