Involvement of Phosphatidylinositol 3-Kinase-Mediated Up-Regulation of IκBα in Anti-Inflammatory Effect of Gemfibrozil in Microglia¹

Microglia are considered as CNS-resident professional macrophages and sensor cells that function as the principal immune effector cells of the CNS responding to any pathological event. Activation of microglia has been implicated in the pathogenesis of a variety of neurodegenerative diseases, including Alzheimer’s disease (AD),³ Parkinson’s disease, Creutzfeld-Jacob disease, HIV-associated dementia (HAD), stroke, and multiple sclerosis (MS) (1). It has been found that activated microglia accumulate at sites of injury or plaques in neurodegenerative CNS (1, 2, 3, 4, 5). Although activated microglia scavenge dead cells from the CNS and secrete different neurotrophic factors for neuronal survival, it is believed that severe activation causes various autoimmune responses leading to neuronal death and brain injury (1, 2, 3, 4, 5). During activation, microglia express various genes related to inflammation, such as proinflammatory cytokines, proinflammatory enzymes, and proinflammatory adhesion molecules (1). Excessive production of these neurotoxic proinflammatory molecules plays an important role in enhancing the degenerative process in the inflamed CNS.

Peroxisome proliferator-activated receptors (PPARs), members of the nuclear hormone receptor superfamily, have been implicated in a variety of human diseases (6). Activation of PPAR-α mainly leads to the induction of a variety of genes such as those coding for the enzymes for β- and ω-oxidation of fatty acids (7). Gemfibrozil, an activator of PPAR-α, has been often prescribed to patients to lower the level of triglycerides (8, 9). This drug decreases the risk of coronary heart disease by increasing the level of high-density lipoprotein cholesterol and decreasing the level of low-density lipoprotein cholesterol (8, 9). In our previous studies (10), we have shown that gemfibrozil markedly inhibits the expression of inducible NO synthase (iNOS) and the production of NO in human astrocytes independent of PPAR-α. The aim of this study was to determine the effect of gemfibrozil on the expression of iNOS and proinflammatory cytokines in microglia and to find out its mode of action. In this study, we demonstrate that gemfibrozil also inhibited the expression of proinflammatory molecules in mouse microglia independent of PPAR-α. Interestingly, gemfibrozil induced the activation of PI3K in microglia and this is the first demonstration of activation of PI3K by any PPAR agonist in any cell type. Interestingly, inhibition of PI3K deleted the anti-inflammatory effect of gemfibrozil while stimulation of PI3K enhanced the same effect establishing a novel role of PI3K in mediating the anti-inflammatory function of gemfibrozil. We also demonstrate that gemfibrozil induced the expression of anti-inflammatory molecule IκBα in microglia through the PI3K pathway.

FBS and DMEM/F-12 were obtained from Invitrogen Life Technologies. LPS (Escherichia coli), gemfibrozil, and fenofibrate were obtained from Sigma-Aldrich. WY-14643 was purchased from Biomol. Wortmannin, LY294002, and Abs against the regulatory subunit of PI3K (p85α) were obtained from Calbiochem. Abs against catalytic subunits of PI3K (p110α and p110β) and chromatin immunoprecipitation (ChIP) grade anti-p65 Abs were obtained from Santa Cruz Biotechnology. [γ-³²P]ATP (3000 Ci/mM) was obtained from PerkinElmer. The dominant-negative mutant of p85α (Δp85α), the constitutively active mutant of p110α/p110β (p110*), and the kinase-dead mutant of p110α/p110β (p110-kd) were provided by Dr. J. R. Raymond (Medical University of South Carolina, Charleston, SC). The expression construct of PPAR-α and the dominant-negative mutant of PPAR-α (ΔPPAR-α) were provided by Dr. S. M. Fischer (University of Texas MD Anderson Cancer Institute, Houston, TX). The superrepressor (SR) construct of IκBα was provided by S. Ghosh (Yale University, New Haven, CT). PPAR-α^−/− mice and littermate controls were purchased from The Jackson Laboratory.

Microglial cells were isolated from mixed glial cultures according to the procedure of Guilian and Baker (11). Animal maintenance and experimental protocols were approved by the Rush University Animal Care Committee. Briefly, mixed glial cells were prepared from 7- to 9-day-old mouse pups. On day 9, the mixed glial cultures were washed three times with DMEM/F-12 and subjected to a shake at 240 rpm for 2 h at 37°C on a rotary shaker. The floating cells were washed and seeded onto plastic tissue-culture flasks and incubated at 37°C for 1 h. The attached cells were removed by trypsinization and seeded on to new plates for further studies. To monitor purity, cells were immunostained with Abs (BD Pharmingen) against Mac-1 surface Ag, a marker for microglia/macrophages. Ninety to 95% of this preparation was found to be positive for Mac-1. For the induction of proinflammatory molecule production, cells were stimulated with LPS in serum-free DMEM/F-12. Mouse BV-2 microglial cells (a gift from V. Bocchini (University of Perugia, Perugia, Italy)) were also maintained and activated by LPS as indicated above.

Synthesis of NO was determined by assay of culture supernatant for nitrite, a stable reaction product of NO with molecular oxygen, using Griess reagent as described earlier (12, 13, 14). Protein was measured by the procedure of Bradford (15).

Cells were stimulated with LPS under serum-free condition and concentrations of TNF-α, IL-1β, and IL-6 were measured in culture supernatants by a high-sensitivity ELISA (BD Pharmingen) according to manufacturer’s instruction as described earlier (16, 17).

After stimulation, cells were lysed with ice-cold lysis buffer containing 1% v/v Nonidet P-40, 100 mM NaCl, 20 mM Tris (pH 7.4), 10 mM iodoacetamide, 10 mM NaF, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl chloride, 1 μg/ml leupeptin, 1 μg/ml antipain, 1 μg/ml aprotinin, and 1 μg/ml pepstatin A. Lysates were incubated at 4°C for 15 min followed by centrifugation at 13,000 × g for 15 min. The supernatant was precleared with protein G-Sepharose beads (Bio-Rad) for 1 h at 4°C followed by the addition of 1 μg/ml p85α mAb. After a 2-h incubation at 4°C, protein G-Sepharose beads were added, and the resulting mixture was further incubated for 1 h at 4°C. The immunoprecipitates were washed twice with lysis buffer, once with PBS, once with 0.5 M LiCl and 100 mM Tris (pH 7.6), once in water, and once in kinase buffer (5 mM MgCl₂, 0.25 mM EDTA, and 20 mM HEPES (pH 7.4)). PI3K activity was determined as described earlier (18, 19) using a lipid mixture of 100 μl of 0.1 mg/ml phosphatidylinositol and 0.1 mg/ml phosphatidylserine dispersed by sonication in 20 mM HEPES (pH 7.0) and 1 mM EDTA. The reaction was initiated by the addition of 20 μCi of [γ-³²P]ATP (3,000 Ci/mM; NEN) and 100 μM ATP and terminated after 15 min by the addition of 80 μl of 1 N HCl and 200 μl of chloroform:methanol (1:1). Phospholipids were separated by thin-layer chromatography and visualized by exposure to iodine vapor and autoradiography (18, 19). Similarly to monitor p110α- and p110β-associated PI3K activity, supernatants were immunoprecipitated with Abs against p110α and p110β followed by the immunocomplex lipid kinase assay as described above.

Class IA PI3K consists of a catalytic subunit (p110) of 110 kDa and a regulatory subunit (p85) of 85 kDa. In the dominant-negative form of p85α, 35 aa in the inter-Src homology 2 (SH2) region from residues 479–513 of wild-type p85α, important for binding the p110α/p110β subunit of PI3K, are deleted, and two other amino acids (Ser-Arg) are inserted in this deleted position. The engineering of the construct and description of the vector driving the expression of the proteins have been published previously (20). In contrast, in the constitutively active mutant of p110α/p110β (p110*), the inter-SH2 domain of p85 is ligated to the NH₂ terminus of p110 whereas in the kinase-deficient mutant of p110α/p110β (p110-kd), the ATP-binding site is mutated (21). Cells plated in 12-well plates were transfected with 0.2–0.25 μg of different plasmids using Lipofectamine Plus (Invitrogen Life Technologies) using the manufacturer’s protocol as described previously (12, 13).

The expression of different proinflammatory molecules was analyzed by semiquantitative RT-PCR using a RT-PCR kit from BD Clontech as described earlier (16, 22). Briefly, total RNA was isolated from stimulated or unstimulated cells by using the Qiagen mini kit followed by digestion with DNase to remove contaminating genomic DNA. Briefly, 1 μg of DNase-digested RNA was reverse transcribed using oligo(dT)_12–18 as primer and MMLV reverse transcriptase (BD Clontech) in a 20-μl reaction mixture. The resulting cDNA was appropriately diluted, and diluted cDNA was amplified using Titanium TaqDNA polymerase and the following primers. The following primers were used to amplify mouse proinflammatory molecules: iNOS (497 bp): sense: 5′-CCC TTC CGA AGT TTC TGG CAG CAG C-3′, antisense: 5′-GGC TGT CAG AGC CTC GTG GCT TTG G-3′; IL-1β (563 bp): sense: 5′-ATG GCA ACT GTT CCT GAA CTC AAC T-3′, antisense: 5′-CAG GAC AGG TAT AGA TTC TTT CCT TT-3′; TNF-α (354 bp): sense: 5′-TTC TGT CTA CTG AAC TTC GGG GTG ATC GGT CC-3′, antisense: 5′-GTA TGA GAT AGC AAA TCG GCT GAC GGT GTG GG-3′; IL-6 (155 bp): sense: 5′-TGG AGT CAC AGA AGG AGT GGC TAA G-3′, antisense: 5′-TCT GAC CAC AGT GAG GAA TGT CCA C-3′; GAPDH (276 bp): sense: 5′-GGT GAA GGT CGG TGT GAA CG-3′, antisense: 5′-TTG GCT CCA CCC TTC AAG TG-3′.

Amplified products were electrophoresed on 1.8% agarose gels and visualized by ethidium bromide staining. GAPDH was used to ascertain that an equivalent amount of cDNA was synthesized from different samples. The relative expression of cytokines or iNOS (cytokines or iNOS/GAPDH) was measured after scanning the bands with a Fluor Chem 8800 Imaging System (Alpha Innotech).

Real-time PCR analysis was performed using the ABI-Prism7700 sequence detection system (Applied Biosystems) as described earlier (16, 22). Briefly, it was performed in a 96-well optical reaction plate (Applied Biosystems) on cDNA equivalent to 50 ng of DNase-digested RNA in a volume of 25 μl, containing 12.5 μl of TaqMan Universal Master mix and optimized concentrations of FAM-labeled probe, forward and reverse primers following manufacturer’s protocol. All primers and FAM-labeled probes for mouse iNOS, cytokines, and GAPDH were obtained from Applied Biosystems. The mRNA expression of iNOS and cytokines was normalized to the level of GAPDH mRNA. Data were processed by the ABI Sequence Detection System 1.6 software and analyzed by ANOVA.

ChIP assays were performed using a kit (Upstate Biotechnology) according to the manufacturer’s protocol. Briefly, 2 × 10⁶ microglial cells preincubated with gemfibrozil for 2 h were stimulated with LPS. After 3 h of stimulation, cells were fixed by adding formaldehyde (1% final concentration), and cross-linked adducts were resuspended and sonicated, resulting in an average chromatin fragment size of 400 bp. ChIP was performed on the cell lysate by overnight incubation at 4°C with 2 μg of Abs against p65 followed by incubation with protein G-agarose (Santa Cruz Biotechnology) for 2 h. The beads were washed and incubated with elution buffer. To reverse the cross-linking and purify the DNA, precipitates were incubated in a 65°C incubator overnight and digested with proteinase K. DNA samples were then purified, precipitated, and precipitates were washed with 75% ethanol, air-dried, and resuspended in Tris-edta buffer. The following primers were used to amplify fragments flanking proximal NF-κB elements in the mouse iNOS promoter: sense: 5′-CAT GAG GAT ACA CCA CAG AG-3′, antisense: 5′-AAG ACC CAA GCG TGA GGA GC-3′.

The following primers were used to amplify fragments flanking distal NF-κB elements in the mouse iNOS promoter: sense: TGC TAG GGG GAT TTT CCC TCT CTC-3′, antisense: 5′-ACC CTG TTC TGA GAA ACA AA-3′; sense: 5′-GAT GTG CTA GGG GGA TTT TCC C-3′; antisense: 5′-TGG GCT AGC CTG GTC TAC AGA G-3′.

The PCRs were repeated by using varying cycle numbers and different amounts of templates to ensure that results were in the linear range of PCR.

Cells plated at 50–60% confluence in 12-well plates were cotransfected with 0.25 μg of pNF-κB-Luc (NF-κB-dependent reporter construct) and 12.5 ng of pRL-TK (a plasmid encoding Renilla luciferase, used as transfection efficiency control; Promega) using Lipofectamine Plus (Invitrogen Life Technologies). After 24 h of transfection, cells were stimulated with different stimuli for 6 h. Firefly and Renilla luciferase activities were analyzed in cell extracts using the Dual Luciferase kit (Promega) in a TD-20/20 Luminometer (Turner Designs) as described earlier (12, 13). Relative luciferase activity of cell extracts was typically represented as the ratio of firefly luciferase value:Renilla luciferase value × 10⁻³.

Mitochondrial activity was measured with the MTT assay (Sigma-Aldrich).

Statistical comparisons were made using one-way ANOVA followed by the Student t test.

Cells were cultured in serum-free medium in the presence LPS. It is evident from Table I that LPS alone markedly induced the production of NO and proinflammatory cytokines (TNF-α, IL-1β, and IL-6) in BV-2 microglial cells. Next, we examined the effect of gemfibrozil, an activator of PPAR-α (23), on LPS-induced production of proinflammatory molecules. Gemfibrozil itself was neither stimulatory nor much inhibitory to NO and cytokine production in control cells. However, gemfibrozil, when added 2 h before the addition of LPS markedly inhibited LPS-induced production of NO, TNF-α, IL-1β, and IL-6 (Table I). Time-course studies show that inhibition was maximum when gemfibrozil was added 2 h before the addition of LPS and the degree of inhibition decreased when gemfibrozil was added either together with LPS or after the addition of LPS (data not shown). Although gemfibrozil at a concentration of 100 μM was not very effective in inhibiting the production of proinflammatory molecules (data not shown), significant inhibition was observed at higher concentrations (200 and 300 μM) of gemfibrozil (Table I). Similar to gemfibrozil, other PPAR-α agonists, such as WY-14643 and fenofibrate, also suppressed the production of nitrite in LPS-stimulated cells at 200 and 300 μM concentrations (Table II). To understand the mechanism of inhibition, we examined the effect of gemfibrozil on mRNA levels of iNOS and proinflammatory cytokines in LPS-stimulated cells. Gemfibrozil dose-dependently inhibited LPS-induced expression of iNOS (49% inhibition at 200 μM gemfibrozil), TNF-α (52% inhibition at 200 μM gemfibrozil), IL-1β (92% inhibition at 200 μM gemfibrozil), and IL-6 (90% inhibition at 200 μM gemfibrozil) mRNAs in BV-2 microglial cells (data not shown). Similarly, at 200 μM concentration, WY-14643 and fenofibrate also attenuated the expression of iNOS and cytokines in LPS-stimulated microglial cells (see Fig. 3, C and D). To examine whether gemfibrozil affected cell viability, microglial cells were incubated with different concentrations (100, 200, and 300 μM) of gemfibrozil under serum-free condition as mentioned above and their viability was determined by the MTT assay. Gemfibrozil at different concentrations used did not decrease the viability of the cells (data not shown). Therefore, inhibition of the expression of proinflammatory molecules by gemfibrozil was not due to any change in viability of the cells.

Because gemfibrozil is a known activator of PPAR-α (6, 23, 24), a member of the nuclear hormone receptor superfamily, we examined whether gemfibrozil inhibited the expression of proinflammatory molecules through the activation of PPAR-α. Therefore, primary microglia were isolated from wild-type and PPAR-α^−/− mice. As evidenced in Fig. 1, LPS induced the mRNA expression of iNOS, TNF-α, IL-1β, and IL-6 in both PPAR-α^+/+ and PPAR-α^−/− microglia. However, gemfibrozil, when added 2 h before the addition of LPS, markedly inhibited LPS-induced expression of proinflammatory molecules in both PPAR-α^+/+ and PPAR-α^−/− microglia (Fig. 1). Similarly, overexpression of either wild-type PPAR-α or dominant-negative PPAR-α (ΔPPAR-α) did not alter the inhibitory effect of gemfibrozil on LPS-induced expression of proinflammatory molecules (iNOS, TNF-α, IL-1β, and IL-6) in mouse BV-2 microglial cells (data not shown). Taken together, these results suggest that PPAR-α is not involved in gemfibrozil-mediated anti-inflammatory effect in microglial cells.

Next, we investigated mechanisms by which gemfibrozil may transduce inhibitory signals for the expression of proinflammatory molecules in microglia. PI3K, a dual protein and lipid kinase, transduces signals for multiple biological processes (25, 26). Class IA PI3K, which is regulated by receptor tyrosine kinases, consists of a heterodimer of a regulatory 85-kDa subunit and a catalytic 110-kDa subunit (p85:p110α/p110β). In contrast, class IB PI3K consists of a dimer of a 101-kDa regulatory subunit and a p110γ catalytic subunit (p101/p110γ). Several recent studies have demonstrated p110γ as a potent mediator of inflammation because the expression of proinflammatory molecules and activation of proinflammatory transcription factor (NF-κB) are reduced in p110γ^−/− cells (27). As a result, either knocking out the p110γ gene or specific inhibition of p110γ by small cell-permeable molecules attenuates the disease process in several inflammatory models (27, 28). In contrast, earlier we have demonstrated that overexpression of constitutively active mutant of p110α/p110β down-regulates the induction of iNOS in astroglial cells (19). Therefore, we were prompted to investigate whether gemfibrozil uses p85α-associated PI3K (p110α/p110β) to inhibit the expression of proinflammatory molecules in microglia. At first, we examined whether gemfibrozil alone was sufficient for the activation of PI3K in microglial cells. We assayed the lipid kinase activity of p85α-associated PI3K. Interestingly, we observed that gemfibrozil induced the activation of p85α-associated PI3K at different time points of treatment as evidenced by lipid kinase activity (Fig. 2, A and B). An ∼>3-fold induction of PI3K was observed at 5 min of stimulation that subsequently peaked at 15 min of stimulation and decreased afterward (Fig. 2, A and B). We compared the magnitude of this activation with insulin, the molecule that is considered to be a prototype inducer of PI3K activation. Although gemfibrozil caused ∼10-fold induction of PI3K after 15 min of stimulation (left panels of Fig. 2, A and B), insulin exhibited an ∼13-fold induction of PI3K at the same time point (right panels of Fig. 2, A and B). Next, we tried to identify the catalytic subunit involved in this lipid kinase activity. The regulatory subunit p85α may associate itself with both p110α and/or p110β. By immunocomplex lipid kinase assay, we have observed that gemfibrozil-induced activation of PI3K was due to p110β but not p110α (Fig. 2, C and D). We also investigated the possibility of whether this increase in PI3K activity was due to any increased expression of PI3K subunits. It is clearly evident from Fig. 2 E that gemfibrozil was unable to increase the mRNA expression of p85α, p110α, and p110β at various time points within 60 min of stimulation nullifying the above possibility.

Because gemfibrozil induced the activation of PI3K, we investigated the possibility of whether gemfibrozil was using this PI3K activity to inhibit the expression of proinflammatory molecules in microglia. We examined the effect of wortmannin and LY294002, inhibitors of PI3K, on the gemfibrozil-mediated anti-inflammatory effect. It is clearly evident from Fig. 3, A and B, that both wortmannin and LY294002 markedly abrogated the inhibitory effect of gemfibrozil on the mRNA expression of iNOS and proinflammatory cytokines in LPS-stimulated BV-2 microglial cells, indicating the dependence of gemfibrozil on PI3K for the anti-inflammatory effect. Next, we investigated whether other PPAR-α agonists (WY-14643 and fenofibrate) also use the PI3K pathway to exert an anti-inflammatory effect. Both WY-14643 and fenofibrate inhibited the expression of iNOS and IL-6 mRNAs in LPS-stimulated BV-2 microglial cells. However, inhibition of PI3K by wortmannin negated this anti-inflammatory effect of both WY-14643 (Fig. 3,C) and fenofibrate (Fig. 3 D), suggesting that agonists of PPAR-α exert their anti-inflammatory effect in microglial cells via the PI3K pathway.

Next, we investigated whether gemfibrozil required PI3K to inhibit the expression of proinflammatory molecules in primary microglia as well. Similar to BV-2 microglial cells, wortmannin was also capable of abrogating the inhibitory effect of gemfibrozil on the expression of iNOS, TNF-α, and IL-6 mRNAs in LPS-stimulated mouse primary microglia (Fig. 4), suggesting the requirement of PI3K in the anti-inflammatory effect of gemfibrozil in primary microglia.

Activated microglia are considered to play an important role in various pathological conditions associated with viral encephalopathy, AD, Parkinson’s disease, HAD, Creutzfeldt-Jakob disease, etc. Because gemfibrozil inhibited LPS-induced expression of proinflammatory molecules in microglia, we were prompted to investigate whether gemfibrozil was also capable of negating the expression of iNOS in microglial cells stimulated with etiological reagents of various neurological disorders. BV-2 microglial cells challenged with fibrillar Aβ peptides (etiological reagent for AD), fibrillar PrP peptides (etiological reagent for prion diseases), dsRNA in the form of poly IC (one of the etiological reagents for viral encephalopathy), HIV-1 Tat (one of the etiological reagents for HAD), and MPP⁺ (Parkinsonian toxin) expressed iNOS mRNA (Fig. 5). However, gemfibrozil knocked down Aβ-, PrP-, poly IC-, Tat-, and MPP⁺-induced microglial iNOS mRNA expression (Fig. 5, A–E). Similar to the regulation of LPS-mediated expression of iNOS, wortmannin also abrogated the inhibitory effect of gemfibrozil on Aβ-, PrP-, poly IC-, Tat-, and MPP⁺-induced microglial expression of iNOS (Fig. 5, A–E), suggesting that gemfibrozil inhibits microglial expression of iNOS induced by etiological reagents of various neurological disorders via a PI3K-sensitive pathway. In contrast, under the same experimental condition, gemfibrozil had no effect on IFN-γ-induced microglial expression of iNOS (Fig. 5 F), suggesting that IFN-γ induces iNOS in microglia via a gemfibrozil-insensitive pathway.

Activation of p85α-associated PI3K (p110β) by gemfibrozil and abrogation of the gemfibrozil-mediated inhibitory effect on LPS-induced expression of proinflammatory molecules in microglia by wortmannin suggests that gemfibrozil possibly exerts its anti-inflammatory effect through the activation of p85α-associated PI3K. To confirm this observation further, we transfected BV-2 glial cells with a dominant-negative mutant of p85α. Expression of a dominant-negative mutant of p85α, in which the inter-SH2 region required for binding of p110α/p110β subunit is disrupted, results in the inhibition of PI3K activity in different cell types including adipocytes and Chinese hamster ovary cells (20, 29). Earlier Pahan et al. (18) have demonstrated inhibition of p85α-associated lipid kinase activity of PI3K in C₆ glial cells by the same dominant-negative mutant of p85α indicating that the overexpressed dominant-negative mutant protein of p85α did not associate with the catalytic subunit of PI3K. As evidenced by semiquantitative RT-PCR analysis in Fig. 6, LPS induced the expression of proinflammatory molecules (iNOS, TNF-α, and IL-6) in empty vector- as well as Δp85α-transfected BV-2 microglial cells. However, gemfibrozil inhibited the mRNA expression of these proinflammatory molecules in empty vector- but not Δp85α-transfected microglial cells (Fig. 6) suggesting again that gemfibrozil exhibits its anti-inflammatory effect in microglia via p85α-associated PI3K.

Because the expression of Δp85α blocked the anti-inflammatory effect of gemfibrozil (Fig. 6) and p85α dimerizes with either p110α or p110β, to confirm this finding further by a different approach, we examined the effect of p110* (constitutively active mutant of p110α/p110β) on the anti-inflammatory effect of gemfibrozil in microglial cells. It has been reported that expression of p110* but not that of p110-kd is sufficient to promote Glut 4 translocation in adipocytes (21). Therefore, to increase the activity of p110α/p110β, mouse microglial cells were transfected with p110*. Earlier, we have demonstrated that expression of the p110* but not that of the p110-kd increases the lipid kinase activity of PI3K in human astroglial cells (19). Semiquantitative RT-PCR analysis in Fig. 7,A and quantitative real-time PCR analysis in Fig. 7,B clearly demonstrate that p110* enhanced the inhibitory effect of gemfibrozil on the expression of proinflammatory molecules compared with empty vector-transfected cells. In contrast, expression of p110-kd blocked this anti-inflammatory effect of gemfibrozil in LPS-stimulated microglial cells (Fig. 7), clearly delineating an essential role of p110β in gemfibrozil-mediated inhibition on proinflammatory molecules.

Next, we investigated mechanisms by which activation of p110β PI3K may limit the expression of proinflammatory molecules in microglia. Because the activation of NF-κB is necessary for the transcription of proinflammatory molecules (12, 13, 30, 31, 32), we investigated the role of NF-κB activation in the expression of iNOS. Mouse iNOS promoter harbors two NF-κB-binding sites: distal (nucleotides −971 to −962) and proximal (nucleotides −85 to −76) (32). At first, we used ChIP analysis to study the recruitment of RelA p65 to each of these two NF-κB-binding sites. After immunoprecipitation of LPS-stimulated microglial chromatin fragments by Abs against p65, we were able to amplify 307-bp fragments flanking the proximal NF-κB element (Fig. 8,A). However, after several attempts, we failed to detect any amplification product spanning the distal NF-κB-binding site (data not shown). These results suggest that LPS induced the recruitment of p65 to the proximal NF-κB-binding site of the mouse iNOS promoter. Therefore, next we examined the effect of gemfibrozil on the recruitment of p65 to the proximal NF-κB-binding site of the iNOS promoter. Consistent to the inhibition of iNOS mRNA expression, gemfibrozil inhibited the recruitment of p65 to the iNOS promoter in LPS-stimulated microglia (Fig. 8,A). In contrast, no amplification product was observed in any of the immunoprecipitates obtained with control IgG (left three lanes of Fig. 8 A) suggesting the specificity of these interactions. These results also suggest that gemfibrozil interferes with the recruitment of NF-κB to the iNOS promoter.

Next, we examined whether NF-κB was also involved in Aβ-, PrP-, poly IC-, Tat-, MPP⁺-, IL-1β-, and IFN-γ-induced microglial iNOS mRNA expression. Similar to LPS, other stimuli (Aβ, PrP, poly IC, Tat, IL-1β, and MPP⁺) markedly induced the transcriptional activity of NF-κB as evident from reporter luciferase activity (Fig. 8,C). In contrast, IFN-γ was unable to induce the activation of NF-κB (Fig. 8,C). We used the SR construct of IκBα to inhibit the function of NF-κB. Expectedly, the SR IκBα markedly blocked LPS-, Aβ-, PrP-, poly IC-, Tat-, IL-1β-, and MPP⁺-induced activation of NF-κB (data not shown). Therefore, by using SR IκBα, we investigated whether NF-κB activation was required for Aβ-, PrP-, poly IC-, Tat-, IL-1β-, and MPP⁺-induced microglial expression of iNOS. It is clear from Fig. 8,B that SR IκBα but not an empty vector was capable of suppressing LPS-, Aβ-, PrP-, poly IC-, Tat-, IL-1β-, and MPP⁺-induced expression of iNOS mRNA. In contrast, SR IκBα had no effect on IFN-γ-induced expression of iNOS mRNA (Fig. 8 B).

Because gemfibrozil inhibited the expression of iNOS and the recruitment of NF-κB to the iNOS promoter in activated microglia, we investigated whether gemfibrozil was capable of attenuating the activation of NF-κB in microglia as well. As evident from Fig. 8,C, gemfibrozil attenuated LPS-, Aβ-, PrP-, poly IC-, Tat-, IL-1β-, and MPP⁺-induced NF-κB-dependent luciferase activity suggesting an inhibitory effect of gemfibrozil on the activation of NF-κB. However, this inhibitory effect of gemfibrozil on neurotoxin-induced NF-κB activation was abrogated by Δp110, a kinase-dead mutant of p110α/p110β, and augmented by p110*, a constitutively active mutant of p110α/p110β, as compared with empty vector-transfected cells (Fig. 8 D). Taken together, these results suggest that all different stimuli of microglial iNOS except IFN-γ induce the expression of iNOS via NF-κB, that gemfibrozil inhibits neurotoxin-induced activation of NF-κB via the PI3K pathway in microglia, and that gemfibrozil is probably unable to interfere with IFN-γ-induced microglial expression of iNOS due to its independency on NF-κB activation.

Because fibrates induce the expression of IκBα capable of arresting the classical NF-κB heterodimer (p65:p50) in the cytoplasm (31), we investigated whether gemfibrozil stimulates the expression of IκBα in microglial cells via a PI3K pathway. As expected, gemfibrozil alone time-dependently induced the expression of IκBα mRNA in BV-2 microglial cells (Fig. 9,A). The increase in IκBα mRNA was visible as early as 15 min with the maximum increase found at 30 min of treatment (Fig. 9,A). Western blot analysis also shows that gemfibrozil induced the expression of IκBα protein at different minutes of stimulation exhibiting the maximum increase at 90 min (Fig. 9,B). Dose-dependent experiment shows that gemfibrozil exhibited maximum increase in IκBα mRNA at a dose of 200 μM or higher (Fig. 9,C). However, interestingly, both wortmannin and LY294002 (inhibitors of PI3K) abrogated gemfibrozil-mediated stimulation of IκBα mRNA (Fig. 9, D and E) suggesting that gemfibrozil induces/stimulates the expression of IκBα via the PI3K pathway. In addition to activating PI3K, gemfibrozil may activate other signaling pathways as well. Therefore, next we investigated whether the activation of PI3K alone was sufficient to induce the expression of IκBα in microglia. It is evident from Fig. 9 F that overexpression of p110* (the constitutively active mutant of p110α/p110β), but not the empty vector, was capable of increasing the mRNA expression of IκBα. In contrast, p110-kd (the kinase-deficient mutant) had no effect on the mRNA expression of IκBα. These results clearly suggest that activation of PI3K is sufficient to induce microglial expression of IκBα.

Common pathological hallmarks of several neurodegenerative diseases include the loss of invaluable neurons associated with or followed by massive activation of microglia (1, 33, 34). Although microglial activation has an important repairing function, once microglia are activated in the neurodegenerating microenvironment, it always goes beyond control and eventually detrimental effects override beneficial effects. Therefore, understanding mechanisms that regulate microglial activation is an important area of investigation that may enhance the possibility of finding a primary or an adjunct therapeutic approach against incurable neurodegenerative disorders. The studies reported in this manuscript clearly demonstrate that gemfibrozil, a commonly used lipid-lowering drug and an activator of PPAR-α, suppresses the expression of proinflammatory molecules (iNOS, TNF-α, IL-1β, and IL-6) in mouse microglia. Because these proinflammatory molecules have been implicated in the pathogenesis of demyelinating and neurodegenerative diseases (3, 5, 34), our results provide a potentially important mechanism whereby activators of PPAR-α may ameliorate neural injury. However, gemfibrozil inhibited the expression of proinflammatory molecules in primary microglia isolated from wild-type as well as PPAR-α^−/− mice. Furthermore, ΔPPAR-α was unable to block the inhibitory effect of gemfibrozil on the induction of proinflammatory molecules in microglial cells. These results suggest that gemfibrozil does not require PPAR-α to display its anti-inflammatory effect in microglia. A recent study by Xu et al. (35) has indicated the possible involvement of PPAR-α in gemfibrozil-mediated inhibition of microglial iNOS. However, that study did not attempt to examine the effect of gemfibrozil in PPAR-α knockout microglia.

Characterization of intracellular pathways that may negatively and positively regulate the expression of proinflammatory molecules in glial cells is an active area of investigation. PI3K is a key signaling molecule implicated in the regulation of a broad array of biological responses including receptor-stimulated mitogenesis, oxidative burst, and cell survival (25, 26). For class IA PI3K, the p85 regulatory subunit acts as an interface by interacting with the insulin receptor substrate-1 through its SH2 domain and thus recruits the p110 catalytic subunit to the cell membrane through its SH2 domain (25, 26). In contrast, for class IB PI3K, the p110γ is activated by the engagement of G protein-coupled receptors. The p110 then catalyzes the reaction to release phosphatidylinositol (3,4,5)-triphosphate as the second messenger using phosphatidylinositol (4,5)-bisphosphate as the substrate and activates downstream signaling molecules like Akt/protein kinase B and p70 ribosomal S6 kinase (25, 26). Earlier Pahan et al. (18) have shown that inhibition of PI3K by either chemical inhibitors, such as wortmannin and LY294002, or a dominant-negative mutant of the regulatory subunit p85α induces/stimulates the expression of iNOS in LPS- or cytokine-stimulated C₆ glial cells and rat primary astrocytes, suggesting that activation of p85α-associated PI3K may transduce an inhibitory signal for the expression of iNOS. Subsequently, we have demonstrated that expression of the catalytically active p110α/p110β subunit inhibits the production of NO and the expression of iNOS in human U373MG astrocytoma cells and primary astrocytes (19). Because these results support the conclusion that p85α-associated PI3K (p110α/p110β) signal transduction pathway is a negative regulator of the expression of iNOS, we were prompted to investigate whether p85α-associated PI3K plays a possible role in anti-inflammatory activity of gemfibrozil in microglia. Several lines of evidence presented in this study clearly support the conclusion that gemfibrozil attenuates the induction of proinflammatory molecules in microglia via PI3K. Our conclusion is based on the following observations. First, gemfibrozil alone induced the activation of p85α-associated PI3K p110β but not p110α in microglia. Second, wortmannin, an inhibitor of PI3K, abrogated the inhibitory effect of gemfibrozil on the expression of proinflammatory molecules in BV-2 microglial cells and primary microglia. Third, overexpression of Δp85α, a dominant-negative mutant of p85α, also blocked the anti-inflammatory function of gemfibrozil. Fourth, overexpression of p110*, a catalytically active mutant of p110α/p110β, enhanced the inhibitory effect of gemfibrozil on the expression of proinflammatory molecules. In contrast, overexpression of p110-kd, a kinase-dead mutant of p110α/p110β, removed this anti-inflammatory effect of gemfibrozil in microglia. Taken together, these studies delineate an absolute requirement of p85α-associated PI3K for the inhibitory effect of gemfibrozil on microglial expression of proinflammatory molecules. However, we do not know mechanisms by which gemfibrozil induces the activation of p85α-associated p110β PI3K in microglia. In general, p85α-associated PI3K is activated via growth factor receptors. Tyrosine phosphorylation of growth factor receptors creates docking sites for binding of p85α through its SH2 domains. Because gemfibrozil is inducing the activation of PI3K within a minute interval, it may not be surprising if gemfibrozil and other fibrate drugs use any of these growth factor receptors to activate PI3K.

Among all the known proinflammatory transcription factors working in concert to transactivate promoters of proinflammatory genes, NF-κB p50:p65 is literally the most important one. The presence of multiple consensus sequences (κB elements) in the promoter region of proinflammatory molecules for the binding of NF-κB and the inhibition of proinflammatory gene expression in human, rat, and mouse glial cells with the inhibition of NF-κB activation (12, 13, 30, 31) establishes an essential role of NF-κB activation in the induction of proinflammatory molecules. In resting cells, the classical p65:p50 heterodimer is arrested in the cytoplasm as an inactive complex by IκBα (31). It has been also demonstrated that newly synthesized IκBα proteins accumulate in cytoplasm as well as in nucleus where it reduces NF-κB binding (36). Several studies have reported that ligands of PPAR inhibit the activation of NF-κB by up-regulating the expression of IκBα (37, 38). We have also observed induction of IκBα mRNA and protein in microglial cells by gemfibrozil. In addition, our results demonstrate a novel mechanism by which gemfibrozil induces/stimulates the expression of IκBα in microglia. Abrogation of gemfibrozil-mediated stimulation of IκBα by inhibitors of PI3K (LY294002 and wortmannin) and up-regulation of IκBα by overexpression of a constitutively active mutant of PI3K alone suggest an important role of PI3K in gemfibrozil-mediated increase in IκBα expression. Therefore, it appears that gemfibrozil-mediated activation p110β PI3K is capable of limiting the expression of proinflammatory molecules in microglia via up-regulation of IκBα. This is interesting because activation of p110γ PI3K usually does the opposite. It induces/stimulates the expression of proinflammatory molecules through enhanced activation of NF-κB, i.e., through enhanced phosphorylation and rapid degradation of IκBα (28).

We extended the study beyond LPS and studied whether gemfibrozil suppressed microglial activation induced by other neurotoxins and etiological reagents of various neurodegenerative disorders via a PI3K pathway. It is important to know that gemfibrozil inhibited microglial expression of iNOS mRNA induced by various neurotoxins, such as Aβ, PrP, poly IC, Tat, IL-1β, and MPP⁺ via a PI3K pathway. Surprisingly, gemfibrozil was unable to suppress IFN-γ-induced expression of iNOS mRNA in microglia. Activation of NF-κB by LPS, Aβ, PrP, poly IC, Tat, IL-1β, and MPP⁺, but not by IFN-γ, and inhibition of LPS-, Aβ-, PrP-, poly IC-, Tat-, IL-1β-, and MPP⁺-, but not IFN-γ- induced expression of iNOS mRNA by SR IκBα suggest that NF-κB is required for LPS-, Aβ-, PrP-, poly IC-, Tat-, IL-1β-, and MPP⁺-, but not IFN-γ-, induced expression of iNOS. It is consistent with a finding by Kleinert et al. (39) that delineates an important role of STAT1α in IFN-γ-induced expression of iNOS. Because gemfibrozil inhibits the activation of NF-κB via PI3K p110β-mediated up-regulation of IκBα, this drug is capable of attenuating the expression of only those proinflammatory molecules whose expression depends on the activation of NF-κB. However, IFN-γ induces the expression of iNOS independent of NF-κB (Fig. 8 B), therefore, gemfibrozil is unable to inhibit IFN-γ-induced microglial expression of iNOS via activation of p110β PI3K.

At present, we do not know the mechanism by which p110β PI3K may transduce signals for enhanced expression of IκBα in microglia. It is known that the promoter of IκBα contains the consensus NF-κB-binding site (40) and once NF-κB is activated, as part of its autoregulation, the transcription of the inhibitory subunit (IκBα) is turned on to suppress its activation. However, gemfibrozil alone does not induce the activation of NF-κB (data not shown) ruling out the possible involvement of NF-κB in transcriptional up-regulation of IκBα. In addition to having NF-κB-binding sites, the promoter of IκBα also houses several consensus sequences for CREB binding (41). Therefore, there is a possibility that gemfibrozil-mediated activated p110β transduces signals for phosphorylation and activation of CREB which in turn is involved in the transcription of the IκBα gene. Experiments are underway in our laboratory to reveal the role of p110β PI3K in gemfibrozil-mediated up-regulation of IκBα in microglia.

Microglia are the sensor cells in the CNS that express proinflammatory molecules in response to any neurotoxic and degenerative insult. Recently, gemfibrozil has been shown to suppress the disease process of experimental allergic encephalomyelitis, an animal model of MS (42), suggesting that gemfibrozil may have a therapeutic effect in this neuroinflammatory disease. Although the in vitro situation of mouse microglia in culture and its treatment with Parkinsonian neurotoxin and etiological reagents of AD, HAD, prion diseases, and viral encephalopathy may not truly resemble the in vivo situation of microglia in the brain of patients with these neurodegenerative disorders, our results identify gemfibrozil as a possible therapeutic agent to suppress microglial activation in neuroinflammatory and neurodegenerative disorders via PI3K p110β-mediated up-regulation of IκBα.

The authors have no financial conflict of interest.