主题：factor

【关键词】  Activation

    Oxidative stress, causing necrotic and apoptotic cell death, is associated with bile acid toxicity. Using liver (HepG2, Hepa1c1c7, and primary human hepatocytes) and intestinal (C2bbe1, a Caco-2 subclone) cells, we demonstrated that toxic bile acids, such as lithocholic acid (LCA) and chenodeoxycholic acid, induced the nuclear factor (erythroid-2 like) factor 2 (Nrf2) target genes, especially the rate-limiting enzyme in glutathione (GSH) biosynthesis [glutamate cysteine ligase modulatory subunit (GCLM) and glutamate cysteine ligase catalytic subunit (GCLC)] and thioredoxin reductase 1. Nrf2 activation and induction of Nrf2 target genes were also evident in vivo in the liver of CD-1 mice treated 7 to 8 h or 4 days with LCA. Silencing of Nrf2 via small-interfering RNA suppressed basal and bile acid-induced mRNA levels of the above-mentioned genes. Consistent with this, overexpression of Nrf2 enhanced, but dominant-negative Nrf2 attenuated, Nrf2 target gene induction by bile acids. The activation of Nrf2-antioxidant responsive element (ARE) transcription machinery by bile acids was confirmed by increased nuclear accumulation of Nrf2, enhanced ARE-reporter activity, and increased Nrf2 binding to ARE. It is noteworthy that Nrf2 silencing increased cell susceptibility to LCA toxicity, as evidenced by reduced cell viability and increased necrosis and apoptosis. Concomitant with GCLC/GCLM induction, cellular GSH was significantly increased in bile acid-treated cells. Cotreatment with N-acetyl-L-cysteine, a GSH precursor, ameliorated LCA toxicity, whereas cotreatment with buthionine sulfoximine, a GSH synthesis blocker, exacerbated it. In summary, this study provides molecular evidence linking bile acid toxicity to oxidative stress. Nrf2 is centrally involved in counteracting such oxidative stress by enhancing adaptive antioxidative response, particularly GSH biosynthesis, and hence cell survival.

    Exposure to excessive bile acids is toxic to the cells, contributing an etiopathological factor to a number of liver and intestinal diseases such as cholestasis and colorectal cancer (Rao et al., 2001; Debruyne et al., 2002). Among the bile acids, lithocholic acid (LCA), a hydrophobic secondary bile acid produced by colonic microflora on chenodeoxycholic acid (CDCA), is the most toxic bile acid, with genotoxic and mutagenesis-enhancing properties (Kawalek et al., 1983; Kozoni et al., 2000). In rodents, it induces intrahepatic cholestasis-like hepatotoxicity (Staudinger et al., 2001), and it promotes chemical-induced colon carcinogenesis (Kozoni et al., 2000). CDCA, the most hydrophobic primary bile acid, is able to cause severe liver injury in species such as rhesus monkey, and it causes mild hepatotoxicity in humans; its chronic administration results in increased colonic production of LCA (Hofmann, 2004).

    The integrity and coordination of efficient hepatic bile flow and intestinal bile extraction are hence critical for species survival. The liver and intestinal cells achieve this through a concerted network involving the nuclear transcription factors, such as farnesoid X receptor (FXR), vitamin D receptor (VDR), retinoid X receptor, liver X receptor, pregnane X receptor (PXR), and/or constitutive androstane receptor. These receptors regulate bile-metabolizing and -conjugation enzymes and bile transporters to prevent excessive accumulation of bile acids (Eloranta and Kullak-Ublick, 2005). Bile acids are regarded as signaling molecules that facilitate synchronization of the above-mentioned regulators in their handling of cellular bile fate. The cross-talk among these receptors is important in maintaining homeostasis of physiological bile extraction, constituting the baseline protection against bile acid toxicity.

    Meanwhile, increased cellular production of reactive oxygen species (ROS) and oxidative stress has been implicated in exposure to toxicological concentrations of bile acids. Bile acid-induced oxidative stress results from induction of membrane permeability transition consequent to mitochondrial toxicity and activation of death receptors (CD95), which subsequently lead to apoptosis, via activation of proapoptotic effectors caspases, and necrosis (Palmeira and Rolo, 2004). Whether there exists any regulator to counteract such oxidative stress and the progression of bile acid toxicity is presently unknown. Because of its important role as an oxidative stress sensor and antiapoptosis factor, we hypothesized that the nuclear factor (erythroid 2-like) factor 2 (Nrf2) may play a central role by enhancing adaptive response and cell survival during exposure to excess bile acids.

    Nrf2, a basic leucine zipper transcription factor that binds to antioxidant responsive element (ARE), is a chief regulator for many antioxidative, cytoprotective genes (Kensler et al., 2007). Among Nrf2 target genes, the glutamate cysteine ligase (GCL), composed of modulatory (GCLM) and catalytic (GCLC) subunits, is the rate-limiting enzyme for cellular biosynthesis of glutathione (GSH), an important intracellular antioxidant in preserving redox balances. Emerging studies have shown that Nrf2 is a multiorgan protector against various toxic reactive insults; among others are chemical carcinogens (Ramos-Gomez et al., 2001) and acetaminophen (Chan et al., 2001). Hence, robust Nrf2 activation in the cell may be a critical adaptive response to overcome oxidative stress-induced disease processes. However, Nrf2 activation is not merely a cellular response to all circumstances of oxidative stress, because exposure to some oxidative stress inducers such as high-dose UVB ray would in turn result in Nrf2 deactivation (Kannan and Jaiswal, 2006). Presently, it is not known whether toxic bile acids could activate Nrf2.

    In this study, we combined in vitro and in vivo approaches to demonstrate that Nrf2 is activated by cytotoxic bile acids, thereby inducing genes, such as GCL and hence GSH biosynthesis, to protect the cells against bile acid toxicity.

    Cell Culture and Chemicals. The human hepatoma-derived HepG2 (American Type Culture Collection, Manassas, VA) and mouse hepatoma-derived Hepa1c1c7 (a gift from Dr. Patricia Harper, The Hospital for Sick Children, Toronto, ON, Canada) were maintained in -MEM with 10% fetal bovine serum (FBS). C2bbe1, a subclone of colon carcinoma Caco-2 that displays a more homogeneous brush-border epithelial-like morphology (American Type Culture Collection), was maintained in Dulbecco's modified Eagle's medium supplemented with 10% FBS, 1.5 g/l sodium bicarbonate, and 10 mg/l holo-transferrin. The human primary hepatocytes were purchased from Celprogen (San Pedro, CA), and they were grown in specially formulated serum-free growth media (Celprogen). Experiments of all secondary cell lines were conducted within 10 cell passages. Treatments were given at 80% confluence for all cell lines except for C2bbe1. Because C2bbe1 cells differentiate to mature colonocytes at confluence, treatments to this cell line were given 2 to 3 days after confluence. All chemicals were purchased from Sigma-Aldrich (St. Louis, MO), unless otherwise indicated. Test bile acids were dissolved in dimethyl sulfoxide (DMSO) [0.2% (v/v)]. Oligonucleotides were synthesized at the Toronto Centre for Applied Genomics (Toronto, ON, Canada) or Integrated DNA Technologies, Inc. (Coralville, IA).

    In Vivo Mouse Experiments. The animal care and experimental procedures were approved by the Animal Care Committee at the University of Toronto and the Hospital for Sick Children (Toronto, ON, Canada). To examine whether Nrf2 target genes may have been modulated during acute exposure to LCA before the onset of symptomatic liver injury, 9- to 12-week-old CD-1 mice (Charles River Canada, Montreal, QC, Canada) were injected i.p. with a single dose of LCA at 125 mg/kg b.wt. dissolved in sterilized DMSO (final amount, <1% b.wt.). Mice were killed 7 to 8 h after the treatment. In a separate experiment aiming to investigate changes in similar genes upon extended treatment with LCA, mice were injected the same dose of LCA dissolved in sterilized corn oil (final amount 2% b.wt.) twice daily for 4 days. This treatment protocol has been used in the past to induce cholestatic liver injury in mice (Staudinger et al., 2001). Mice were killed 16 h after the last dosing. The use of corn oil as a solvent for LCA in the extended treatment protocol was to avoid the possible toxicity with chronic exposure to DMSO. At necropsy, portions of their livers were sampled in RNAlater reagent (Invitrogen, Carlsbad, CA) and neutral-buffered 10% formalin for mRNA and histological analyses, respectively. Nuclear protein extraction of chilled livers was carried out using Nuclear Extraction kit (Panomics, Fremont, CA). Sera of mice were collected for analysis of liver function/injury markers: total bilirubin (TBL), alanine aminotransferase (ALT), aspartate aminotransferase (AST), and -glutamyl transferase using established automated methods (Department of Pediatric Laboratory Medicine, The Hospital for Sick Children).

    cDNA Synthesis and Quantitative Reverse-Transcription Polymerase Chain Reaction. Total RNA was isolated with RNeasy kit (QIAGEN, Valencia, CA) and reverse-transcribed into cDNA using random hexamers and Moloney murine leukemia virus or SuperScript II reverse transcriptase (Invitrogen). Aliquots of cDNA equivalent to 100 ng of RNA were used for real-time PCR performed on Applied Biosystems (Foster City, CA) 7500 Real-Time PCR system or Prism 7700 Sequence Detection system with reaction mode set at 50°C for 2 min, 95°C for 20 s, followed by 40 cycles of 95°C for 15 s and 56 or 60°C for 1 min. The primers for ribosomal 18S, -actin, tata-box binding protein, glyceraldehyde-3-phosphate dehydrogenase, GCLM, GCLC, and NAD(P)H quinone oxidoreductase 1 (NQO1) were purchased from predesigned and preoptimized Taqman primer-probe sets (Assay-on-Demand Gene Expression probe; Applied Biosystems), whereas custom-made primers for SYBR Green real-time PCR detection were used for the other gene transcripts (primer sequences available upon request). To ensure specificity, primer pairs were designed to span across two neighboring exons and detection of a single peak in dissociation curve analysis. The Ct method (Livak and Schmittgen, 2001) was used to quantify the amplification-fold difference between treatment and vehicle-treated control groups, with the Ct value of target genes being adjusted to individual housekeeping gene (glyceraldehyde-3-phosphate dehydrogenase, -actin, tata-box binding protein, and/or 18S), whichever expression was not affected by treatment protocols. Measurements were done in duplicate or triplicate with variability <0.5 Ct.

    Immunoblotting. Whole cell lysate was prepared in radioimmunoprecipitation assay buffer with protease inhibitor cocktail (Roche Applied Science, Indianapolis, IN). Ten to 30 µg of protein was dissolved in 4 to 12% bis-tris gel (NuPage Novex gel system; Invitrogen) and then transferred onto a nitrocellulose membrane (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK). Primary antibodies (working concentration) used were as follows: rabbit polyclonal anti-GCLC antibody-1 (1:2000) (NeoMarkers, Fremont, CA), rabbit antiserum against GCLM (1:3000) (custom-made; Alpha Diagnostic, San Antonio, TX; see below), rabbit polyclonal anti-Nrf2 c-20 (1:750) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), rabbit polyclonal anti-thioredoxin reductase 1 (TRx1) (1:3000) (Abcam Inc., Cambridge, MA), mouse monoclonal anti--actin (1:10,000) (Sigma-Aldrich), and goat polyclonal anti-lamin B c-20 (1:200) (Santa Cruz Biotechnology, Inc.). Based on analyses of hydrophilicity, antigenicity, accessibility, and sequence homology with other related proteins, an antiserum against a peptide (amino acids 29–45) of human GCLM was raised in rabbits. The immunogenicity and specificity were checked by enzyme-linked immunosorbent assay, and its ability to detect an 28-kDa protein (predicted size of GCLM) with reactivity halted after preabsorption of the antibody in excess immunogen. To ensure equal loading for whole cell lysate and nuclear protein, -actin and lamin B, respectively, were probed on the same stripped blot membranes after being used for detecting target proteins.

    RNA Interference. A combo of four gene-specific small-interfering RNAs (siRNA) against human Nrf2 (NM_006164) was used (Dharmacon SMARTpool siRNA reagent; Dharmacon RNA Technologies, Lafayette, CO). Overnight-seeded HepG2 and C2bbE1 cells at 40 and 60% confluence, respectively, were transfected for 48 h with 50 nM siRNA against Nrf2 (siNrf2) or equal molar mismatched siRNA controls (siCtr). These siRNAs were earlier complexed with liposome carrier Dharmafect I (Dharmacon RNA Technologies) at 0.2 µl/nM siRNA concentration in serum-free Opti-MEM (Invitrogen). Under this condition, the transfected cells after 48 h looked normal morphologically, and they did not differ from untransfected cells in cell viability and mRNA levels of inflammatory marker interleukin-6 (data not shown). Treatments with bile acids were then followed for 16 to 18 h. To ensure achieving functional and specific silencing, the mRNA levels of Nrf2, known Nrf2 target genes, and homologous subtypes Nrf1 and Nrf3 were compared between siNrf2 and siCtr groups before and after treatments in all experiments.

    Plasmid Constructs. The expression vectors for Nrf2 (pEF_Nrf2), dominant-negative Nrf2 (pEF_DNrf2), and empty vector (pEF) were kindly provided by Dr. Jawed Alam (Ochsner Clinic Foundation, New Orleans, LA). To make an ARE-reporter construct (pGL3_ARE), a DNA duplex (CGGGGTACCGCCCGCACAAAGCGCTGAGTCACGGGGAGGCAGATCTTCC) (core ARE is underlined; –3595/–3625 of hGCLC gene) containing the indispensable ARE motif of GCLC (–3604/–3614) (Mulcahy et al., 1997) with KpnI/BglII at 5' and 3' ends, respectively, was constructed by annealing two polyacrylamide gel electrophoresis-purified complementary oligonucleotides. This insert was ligated to similar restriction enzyme sites of pGL3 luciferase reporter plasmid with simian virus 40 promoter (Promega, Madison, WI). A similar reporter construct has been successfully used previously (Mulcahy et al., 1997). The responsiveness and robustness of our ARE reporter to Nrf2 transactivation was confirmed by testing of a panel of Nrf2 activators such as tert-butylhydroquinone (BHQ), lipoic acid, and diethyl maleate (data not shown), as well as cotransfection with Nrf2 and dominant-negative Nrf2 expression vectors. To ensure specificity, a mutant ARE reporter construct (pGL3_mARE) introducing three point mutations on ARE was cloned by PCR-mediated site-directed mutagenesis using Pfu Turbo DNA polymerase (Stratagene, La Jolla, CA) with complementary primers 5'-AGCtaTGAGgCACGGGGAGGCAG-3' (underlined sequence is core ARE; lowercases are mutated nucleotides) on the template pGL3_ARE. The PCR condition was 95°C for 5 min followed by 22 cycles of 95°C for 15 s, 55°C for 30 s, 68°C for 10 min, and a final extension of 68°C for 10 min. The template was then digested by DpnI, and mutant clones were transformed in XL-1 blue competent cells (Stratagene). Successful insertion and mutation introduction were confirmed by sequencing. The cDNA clone of human Na(+)-taurocholate cotransporting polypeptide (NTCP) (Origene Technologies, Rockville, MD) was subcloned into the NotI site of pTarget expression vector (Promega), and it was stably transfected into HepG2. Stable clones transfected with NTCP or empty vector (pTarget) were selected using 500 µg/ml G-418–supplemented growth media.

    Transfection, Reporter Assays, and Overexpression Studies. HepG2 cells at 50% confluence were transiently transfected overnight with 0.1 µg of the firefly luciferease reporter pGL3_ARE or pGL3_mARE, 0.02 µg of the Renilla reniformis luciferase control reporter pRL-TK with or without cotransfection with 0.2 µg of expression vectors using Lipofectamine 2000 (Invitrogen) as transfection carrier. Treatments with bile acids were then carried out for another 16 to 18 h in all experiments, unless otherwise stated. Conjugated bile acid treatments [glycocholic acid (GCA) and glycochenodeoxycholic acid (GCDCA)] were done on NTCP-transfected HepG2. Luciferase activities of the cell extracts were determined with the Dual-Luciferase Reporter Assay system (Promega). Relative luciferase activity (relative light unit) was calculated from firefly luciferase values normalized to those of the control R. reniformis luciferase, and activity is expressed as ratios to vehicle-treated empty pGL3 promoter construct, and, if any, cotransfected expression vector. All experiments were done in triplicate, and they were repeated at least twice. For overexpression studies, Hepa1c1c7 cells at 50% confluence in T25 flasks were transfected with 3 µg of Nrf2 or dominant-negative Nrf2 expression vector for 24 h, followed by treatments with bile acids for another 20 to 22 h.

    Quantitative Chromatin Immunoprecipitation. The assay was performed using the ChIP assay kit (Upstate, Charlottesville, VA) with slight modifications. After 6 h of treatments, chromatin protein-DNA of HepG2 cells was fixed (cross-linked) in neutral-buffered 1% formaldehyde at room temperature for 10 min. Further fixation was stopped by 125 mM glycine buffer. The DNA was sheared by sonication on ice into fragments of  500 bp. An aliquot (one fourth) of sample supernatant was saved as input DNA for later PCR analysis. After preclearing with protein A agarose beads, supernatants were incubated with a ChIP-graded anti-Nrf2 antibody (1:250; Santa Cruz Biotechnology, Inc.) in rotation at 4°C overnight. To control for nonspecific binding of antibody used, an equal amount of the host antibody against an irrelevant protein (rabbit polyclonal anti-CYP1A1) from the same manufacturer (Santa Cruz Biotechnology, Inc.) was included in a separate batch of control supernatants and followed through the remaining protocols. Antibody-chromatin complexes were collected by salmon sperm DNA/protein A beads. DNA was released from cross-linked complexes with proteinase K at 65°C for 4 h followed by 72°C for 10 min. DNA was then extracted and eluted with 120 µl of Tris, pH 8.0, buffer using the DNeasy kit (QIAGEN), and the contaminant RNA was cleaved with RNase A (Invitrogen). For detection of the ARE of GCLM (–56/–66) (Erickson et al., 2002) and of GCLC (–3604/–3614) (Mulcahy et al., 1997) by real-time PCR, the primer sets and Taqman probe (5'-5-carboxyfluorescein, 3'-5-carboxytetramethylrhodamine) were designed by PrimerQuest software (Integrated DNA Technologies, Inc.), which amplify 5'-region exactly on the core ARE. The primers for detecting the ARE of GCLM were as follows: sense, 5'-CGCGGGATGAGTAACGGT-3'; antisense, 5'-GGGAGAGCTGATTCCAAACTGA-3'; and probe, 5'-ACGAAGCACTTTCTCGGCTACGAT-3', which amplify a 79-bp product (–33/–112). For probing the ARE of GCLC, the primers used were sense, 5'-GGACTGAGACTTTGCCCTAAGAAG-3'; antisense, 5'-GCGCAGTTGTTGTGATACAG-3'; and probe, 5'-CGCACAAAGCGCTGAGTCAC-3', which amplify a 160-bp product (–3479/–3609). Quantitation of NRF2 bound to these AREs after the treatments was carried out on 5% of DNA eluates with qPCR analysis similar to that for the mRNA, except that the Ct value of amplicon from each sample's input DNA was used as normalization control as described previously (Beresford and Boss, 2001).

    Cytotoxicity, Necrosis, and Apoptosis. For cytotoxicity analysis, a nontoxic assay, namely, Alamar Blue (BioSource, Nivelles, Belgium), was used. This assay uses a fluorometric indicator to measure the chemical reduction of cell medium, which correlates directly to the metabolic activity of viable cells. The working assay medium [10% (v/v) Alamar Blue in -MEM, 2% FBS, and 1% penicillin/streptomycin, 37°C] was first incubated with cells seeded on 24-well culture plate before treatment to obtain baseline/pretreatment values. The measurement was made at excitation/emission/cut-off  = 540/590/570 nm after 1 h of incubation with the assay medium at 37°C. Immediately after the measurement, the cells were rinsed with prewarmed PBS followed by the treatments. At various time points, treatment media were replaced with fresh assay media to allow for a continuous monitoring of cell viability. The fluorescent unit of each treatment and control was expressed as percentage of change relative to individual baseline/pretreatment value.

    To determine necrosis, cellular release of lactate dehydrogenase (LDH) into treatment media was measured with an LDH detection kit (Roche Applied Science). To control for cell mass and spontaneous release of LDH by viable cells into media, the ratio of LDH activity in the medium to the cells (cell lysate) was determined and then subtracted from those of the vehicle-treated controls. The measurement was made colorimetrically at  = 490 nm. The intra-assay variability of duplicate determinations was 2.2.

    To assess apoptosis, the cellular caspases activity was measured using the rhodamine 110-conjugated substrate N-benzyloxycarbonyl-Asp-Glu-Val-Asp (Invitrogen). Although traditionally known to detect caspase-3 activity, recent analysis by the manufacturer showed that this substrate is also a target of multiple caspases such as 6, 7, 8, and 10. The caspases activity of cell lysate was quantitated at excitation/emission  = 496/520 nm, and it was normalized to individual protein concentration measured by Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, CA).

    Total GSH Quantitation. Cellular GSH was quantitated with a GSH assay kit (Cayman Chemical, Ann Arbor, MI) based on an established GSH recycling enzymatic method (Tietze, 1969). After 24-h treatment with bile acids, HepG2 cells were lysed in ice-cold MES buffer after a quick freeze-thaw cycle and then deproteinized by 0.5 g/ml metaphosphoric acid. An aliquot of each sample was saved before deproteinization for determining protein content. The total GSH of deproteinized cell supernatants was measured against an oxidized glutathione standard curve according to the manufacturer's instruction. The measurement unit was expressed as nanomoles per milligram of protein per minute. The intra-assay variability of duplicate determinations for all samples repeated in four experiments was 3.1.

    Statistical Analysis. Statistical tests were conducted using SigmaStat 3.1 (Systat Software, Inc., Point Richmond, CA) or SPSS 10.1 (SPSS Inc., Chicago, IL). Normality and equal variance tests were first carried out to guide subsequent statistical analyses. Multiple group comparisons were carried out by one-way ANOVA (parametric) or one-way ANOVA on ranks (nonparametric). Once statistical significance was attained (p < 0.05), the Dunnett's (parametric) or Dunn's (nonparametric) test comparing between treatment and control groups was initiated. Comparisons between two groups on single variable were accomplished by Student's independent t test (parametric) or Mann-Whitney U test (nonparametric). Difference with p < 0.05 was considered statistically significant.

    Induction of Nrf2 Genes by Bile Acids in Liver and Intestinal Cells. A dose-response increase in mRNA of GCLM and GCLC after LCA and/or CDCA treatment was noted for HepG2 and C2bbe1 cells, with a significant >4-fold induction at 50 µM LCA and 100 µM CDCA (Fig. 1a). Peak inductions of GCL subunit genes occurred at 50 to 75 µM LCA and at 100 to 150 µM CDCA. Increased bile acid concentrations, i.e., LCA 100 µM or CDCA 250 µM, resulted in markedly cell death and reduced inductions of GCL genes at 24 h of treatment (data not shown). Significantly higher GCL gene transcripts, although at a lower magnitude (2–4-fold induction), were also noticed for the primary human hepatocytes treated with both bile acids (Fig. 1a). So were similar treatments given to the mouse hepatoma Hepa1c1c7 (data not shown). In C2bbe1 cells, CDCA treatment (100 µM) caused a modest increase in GCL genes (2-fold), whereas treatment with deoxycholic acid (DCA) (100 µM), a secondary bile acid often associated with toxicity and carcinogenesis in colonic cells, resulted in comparable GCL inductions to those of LCA treatment (data not shown).

    Fig. 1. Induction of Nrf2 target genes by bile acids. a, mRNA levels of GCLM (white bar) and GCLC (black bar) in HepG2 and C2bbe1 cells (left) after 24-h treatment with increasing doses of LCA or CDCA: [LCA] = 6.25, 12.5, 25, 50, and 75 µM and [CDCA] = 25, 50, 75, 100, and 150 µM. Right, human primary hepatocytes (passages 4–6) were treated with 75 µM LCA or 100 µM CDCA for 24 h. y-axis, -fold change versus vehicle-treated controls. *, p < 0.05, significantly different from vehicle-treated control by one-way ANOVA followed by post hoc test for HepG2 and C2bbel or **, p < 0.01 by t test for primary hepatocyte. Mean and S.E.M. (n = 3–6). b, mRNA levels of other known Nrf2 target genes after 24-h treatment with bile acids in HepG2, C2bbe1, and primary hepatocytes: [LCA] = 75 µM and [CDCA] = 100 µM. Mean and S.E.M. (n = 3–6).c, representative immunoblots of protein lysate (10 µg for HepG2; 30 µg for C2bbe1) probed for Nrf2 target genes after 24-h treatment with bile acids. [LCA] = 70 µM and [CDCA] = 100 µM. d, cell viability (by Alamar Blue) and apoptotic marker (caspases activity) measured across increasing concentrations of LCA treatment in HepG2. Viability was measured at 24 h, whereas caspases activity was measured at 6 h of LCA treatment. Note that the induction of GCL genes peaked at 60 to 80 µM LCA (shown by closed inverted triangles) during which mild cellular toxicity began to occur. Representative results from four determinations are shown.

    Furthermore, mRNA of Nrf2 and a panel of known Nrf2 target genes such as NQO1, TRx1, ferritin light subunit, and heme oxygenase I were also simultaneously induced 2-fold by bile acids in all test cells (Fig. 1b). Particularly, TRx1, an important seleno-enzyme in cellular thiol and redox maintenance, was increased >4-fold in HepG2 and C2bbe1. Note that glutathione transferase P1, which was induced by bile acids in HepG2 and C2bbe1, was not evident in primary hepatocytes (Fig. 1b). Instead, another GST subtype, glutathione transferase A, was increased by bile acids to 2-fold in primary hepatocytes (data not shown). This disparity suggests possible cell type specificity in regulation of GSTs by bile acids.

    Increased protein levels corresponding to mRNA induction were also noted (Fig. 1c). The apoptosis marker (caspases activity) and cell viability analyses showed that a mild toxicity began to occur in HepG2 cells at 60 to 80 µM LCA treatment, followed by a precipitous increase in cell death and caspase activity at >80 µM (Fig. 1d). It is noteworthy that the induction of GCL subunits and other antioxidative genes peaked in the range of LCA (60–80 µM) during which HepG2 began to experience mild toxicity. These findings suggest that induction of the cytoprotective genes may represent an adaptive cell defense mechanism against LCA toxicity.

    In Vivo Activation of Nrf2 Target Genes. Acute administration (7–8 h) of LCA to mice at a dose known to induce cholestatic liver injury (Staudinger et al., 2001) resulted in Nrf2 accumulation in the nuclei, a signature event of Nrf2 activation (Fig. 2a). This phenomenon coincided with significant inductions of Nrf2 target genes (Fig. 2b, top) found to be increased in the in vitro studies. It is noteworthy that the increase of TRx1 transcripts rose to 50-fold at acute exposure to LCA, implying a possible critical role of this enzyme in early toxicity of LCA. At this shorter exposure to LCA, however, analysis of serum liver function and cholestatic markers (ALT, AST, -glutamyl transferase, and TBL) as well as liver histology did not indicate liver dysfunction or pathological changes (data not shown).

    Fig. 2. Activation of Nrf2 in mice treated with LCA. a, representative immunoblots of nuclear fractions (30 µg) probed against Nrf2 in the liver of mice after 7- to 8-h treatment with cholestatic LCA (125 mg/kg b.wt.). Lamin B was used as loading control for nuclear protein, whereas -actin was probed to show unintentional contamination of cytosolic proteins in nuclear fraction preparation. Note that the increased nuclear Nrf2 cannot be explained by inclusion of contaminant cytosolic proteins. b, mRNA levels of Nrf2 target genes upon LCA treatment for 7 to 8 h (top) or for 4 days (bottom) in the liver of mice. Because there were differences in the basal gene expression of Nrf2 target genes between sexes, comparisons of all target genes between treated and untreated groups were adjusted for sex. Induction of antioxidative genes by LCA, however, occurred to both sexes. Significantly different from vehicle-treated controls by t test. *, p < 0.05; **, p < 0.01. Mean and S.E.M. (n = 5–9 for 7- to 8-h treatment; n = 3 for 4-day treatment).

    With prolonged LCA treatment during which symptomatic liver injury (elevated ALT, AST, and TBL, and liver necrosis in histological analysis; data not shown) already occurred, induction of Nrf2 target genes, such as GCL subunit gene transcripts, was found to sustain compared with those treated acutely with LCA (Fig. 2b, bottom). Nqo1 was increased with prolonged treatment, whereas TRx1 induction was subdued. Consistent with the observations from primary human hepatocytes,  class of mouse Gst (Gsta1/2), rather than glutathione transferase P1, was found to be induced by LCA, with 10-fold induction at 4 days of treatment (Fig. 2b, bottom). Treatment with vehicle alone did not differ in mRNA of genes under study compared with untreated animals (data not shown).

    Involvement of Nrf2 and Activation of Nrf2-ARE Transcription Machinery. To examine whether Nrf2 participated in the preceding gene activations, we silenced Nrf2 of HepG2 and C2bbe1 via siRNA. This resulted in significant reductions of >60% in Nrf2 mRNA and protein without interfering with other homologous Nrf subtypes (Fig. 3, a and b). Nrf2 silencing significantly decreased the basal levels of known Nrf2 target genes (Fig. 3, c–e), an observation similar to that seen in in vivo Nrf2 knockout mice (Lee et al., 2005). In addition, the induction of GCLM, GCLC, and other Nrf2 target genes by bile acids in HepG2 (Fig. 3, c and e) and C2bbe1 (Fig. 3d) has been mitigated. Comparable reduction in inducible expressions of Gclm occurred with transfection of dominant-negative Nrf2 in Hepa1c1c7 cells, consistent with the enhanced gene induction with Nrf2 overexpression (Fig. 3f). Similar observations were noted for other Nrf2 target genes such as Gclc and Nqo1 (data not shown).

    Fig. 3. Involvement of Nrf2 in induction of glutamate cysteine ligase subunits by bile acids. a, mRNA levels of all Nrf subtypes after 72-h treatment with siRNA. y-axis, -fold change versus vehicle-treated cells transfected with siCtr. *, p < 0.001, significant difference between siRNA groups by t test. Mean and S.E.M. (n = 3–5). b, representative immunoblots (20 µg of protein lysate) of HepG2 after 48-h treatment with siRNA against Nrf2. c, basal (treated with vehicle DMSO) and inducible (treated with 70 µM LCA or 100 µM CDCA) gene transcripts of GCLM and GCLC in HepG2 (mean and S.E.M.; n = 3–5). d, basal and inducible levels of GCLM and GCLC gene transcripts in C2bbe1 treated with vehicle or 70 µM LCA. *, p < 0.01, significant difference between siCtr and siNrf2 with or without treatment by t test. Mean and S.E.M. (n = 3–5). e, basal and inducible expressions of other Nrf2 target genes in HepG2. Mean ± S.E.M. f, mRNA levels of GCLM induced by 70 µM LCA or 100 µM LCA in Hepa1c1c7 transfected with empty vector (pEF), Nrf2 (pEF_Nrf2), or dominant-negative Nrf2 (pEF_DNrf2) expression vector. y-axis, -fold change versus vehicle-treated cells transfected with empty vector pEF. *, p < 0.05, significantly different from LCA-treated cells transfected with pEF control by t test. Mean and S.E.M. (n = 4).

    To verify that there was an activation of Nrf2-ARE transcription machinery with exposure to toxic bile acids, we extracted the nuclear proteins of bile-acid-treated HepG2 over different time points across 24 h. Translocation of cytosolic Nrf2 to nucleus represents the prerequisite event of receptor activation. Nrf2 started to be enriched in cell nuclei within 1 to 3 h of bile acid treatments, and it was sustained through 24 h, with CDCA-treated cells showing reduced Nrf2 translocation events with longer time of exposure (24 h) (Fig. 4a). Furthermore, various bile acids were found to increase the activity of an ARE-reporter assay in a dose-dependent manner, suggesting that these bile acids were capable of inducing Nrf2 transactivation (Fig. 4b). The magnitude of luciferase activity of the highest concentration of test bile acids was compatible with those of treatments with antioxidants tert-BHQ (100 µM) and -lipoic acid (200 µM) (data not shown), denoting that bile acids are equally potent Nrf2 activators. Note that there was an 8-fold increase in reporter activity with vehicle DMSO treatment compared with that of the empty vector harboring only the simian virus 40 promoter. This suggests the existence of a strong constitutive Nrf2-ARE transactivational activity in HepG2 cells, an observation in line with the persistent oxidative stress observed in many cancerous cell lines (Brown and Bicknell, 2001). HepG2 cells are deficient in conjugated bile acid transporters such as NTCP, which leads to its resistance to conjugated bile acid-induced oxidative stress (Kullak-Ublick et al., 1996). Transfection of NTCP expression vector hence restores, to some degree, its sensitiveness. In this study, we found that GCDCA, a known cholestatic conjugated bile acid, significantly induced the ARE reporter. This suggests that activation of Nrf2-ARE may be crucial to counteract the toxicity of GCDCA as reported previously (Dent et al., 2005). The potency of bile acids in activating this reporter based upon molarities was LCA > CDCA  DCA > GCDCA  ursodeoxycholic acid (UDCA) > cholic acid (CA) > GCA. This order is in consensus with the toxicity profile of these bile acids, particularly in terms of their ability to produce oxidative stress (Krähenbühl et al., 1994). Overexpression of Nrf2 further enhanced the reporter activity by bile acids, whereas coexpression of dominant-negative Nrf2 attenuated the activity, and a mutant ARE construct was completely uninducible by bile acids (Fig. 4c). Using the quantitative ChIP assay, we found an increased Nrf2 occupancy to the AREs of both GCL subunits in the native cell context upon 6-h treatment with bile acids (Fig. 4d). Taken together, our data suggest that activation of the Nrf2-ARE machinery underlies induction of Nrf2-target genes by toxicological concentrations of bile acids.

    Fig. 4. Bile acids activate Nrf2 transcription machinery. a, representative immunoblots of nuclear fraction (10 µg) extracted from HepG2 treated with bile acids (LCA, 70 µM; CDCA, 100 µM) at indicated time points over 24 h. Lamin B was used as equal loading control for nuclear protein, whereas -actin was probed to show possible contaminant cytosolic proteins in nuclear fraction preparation. b, ARE-reporter (luciferase) activity of HepG2 treated with increasing doses of various bile acids for 16 to 18 h. Abbreviations (doses): LCA (50, 70, and 90 µM), CDCA (50, 100, and 150 µM), CA (150, 200, and 400 µM), DCA (50, 100, and 150 µM), UDCA (50, 100, and 200 µM), GCA (200, 400, and 800 µM), and GCDCA (100, 200, and 400 µM). #, GCA and GCDCA were tested in NTCP-transfected HepG2 for 6 h. y-axis, -fold change in the ratio of luciferase activity (relative luciferase unit) (see Materials and Methods for detail) from those transfected with basic pGL3 promoter construct and treated with vehicle DMSO. *, p < 0.05, significantly different from DMSO-treated pGL3_ARE by one-way ANOVA followed by post hoc test. Mean and S.E.M. (n = 2–4). c, ARE-reporter (pGL3_ARE) activity with coexpression of Nrf2 or dominant-negative Nrf2, and mutant ARE-reporter (pGL3_mARE) activity in HepG2 treated with bile acids (70 µM LCA; 100 µM CDCA). y-axis, -fold change in the ratio of luciferase activity (relative luciferase unit) from those transfected with basic pGL3 promoter construct and/or respective expression vector, and treated with DMSO. *, p < 0.05, significant difference between vehicle control and bile acids by one-way ANOVA followed by post hoc test. Mean and S.E.M. (n = 3–6). d, ChIP analysis examining Nrf2 occupancy to AREs of both GCLM and GCLC genes upon 6-h treatment with bile acids (70 µM LCA; 100 µM CDCA) in HepG2. BHQ (200 µM), known to transcriptionally activate GCLM and GCLC, was included as positive control. Negligible detection from samples incubated with host IgG (anti-CYP1A1) ruled out contribution of nonspecific binding from antibody. *, p < 0.05, significantly different from controls (t = 0) and vehicle treatment by t test. Mean and S.E.M. of triplicate determinations of representative experiments.

    Protective Role of Nrf2 in Bile Acid Toxicity. To directly investigate the role of Nrf2 in protection against toxic bile acids, we first silenced Nrf2 of HepG2 via RNA interference upon which the cells were subjected to toxic LCA challenges. Nrf2 knockdown increased cell susceptibility to toxic LCA, with a significantly decreased cell viability starting from 4 h of treatment (Fig. 5a). Without LCA challenge, Nrf2-knockdown cells did not differ in cell viability from those treated with siCtr (data not shown). Significant protective effects of Nrf2 against LCA toxicity was also observed in C2bbe1 cells, and in HepG2 with 300 µM CDCA (data not shown). To investigate which route of cell death was particularly involved in protection by Nrf2, established markers of necrosis and apoptosis were examined. LCA at 90 µM was used because at this dose, both apoptosis and necrosis were found to simultaneously occur. Necrotic event, as determined by LDH released into the culture media, remained constantly higher in Nrf2 knockdown cells than those of siCtr starting from 2 h of LCA treatment (Fig. 5b). Nrf2 silencing alone did not affect the cellular release of LDH (data not shown). Likewise, in the assessment of apoptosis, Nrf2-knockdown cells exhibited much higher and prolonged elevation of caspases activity than did siRNA control-treated cells upon LCA treatment (Fig. 5c). Silencing of Nrf2 alone did not result in increased basal caspases activity.

    Fig. 5. Nrf2 is a cellular protector against LCA-induced toxicity. a, cell viability of HepG2 after knockdown of Nrf2 via siRNA and treatment with toxic LCA (100 and 120 µM). Viability was measured at indicated time points over 24 h. Fluorescent values of each LCA-treated siRNA group was first subtracted by the average values of individual siRNA group treated with vehicle DMSO, and they are expressed as percentage of change to baseline/pretreatment values. *, p < 0.01, significant difference between siCtr and siNrf2 groups with t test. Representative results; mean ± S.E.M. of four determinations. b, analysis of LDH release ratio, a marker of necrotic event or cell injury, upon LCA challenge (90 µM) in HepG2 after Nrf2 silencing. Values of LCA treatment were subtracted by the average values of vehicle treatment for each siRNA group. Representative results were presented; mean of duplicate determinations. c, analysis of caspases activity upon triggered by LCA toxicity (90 µM) in HepG2 knockdown of Nrf2. Representative results of duplicate determinations were shown.

    Role of GSH in Resisting LCA Toxicity. The increase in GCLM and GCLC, the rate-limiting enzyme in GSH biosynthesis, observed in earlier experiments after LCA (75 µM) or CDCA (100 µM) treatment was accompanied by a significant increase by >4-fold in cellular GSH levels at 24 h (Fig. 6a). This increase was comparable with treatment with 200 µM -lipoic acid, a GSH inducer. To determine whether the induced cellular GSH is a protective mechanism against toxic bile acid, we cotreated HepG2 with a toxic dose of LCA and a GSH biosynthesis blocker, buthionine sulfoximine (BSO), which inhibits the activity of GCL subunits and blocks GSH biosynthesis. BSO treatment together with toxic LCA decreased cell resistance toward LCA exposure with more apparent effects in late treatment (24 h) (Fig. 6b). Furthermore, depletion of cellular GSH by pretreatment with BSO before LCA challenge markedly lifted cell resistance with a drastic drop in cell viability within 4 h of treatment. Conversely, toxic LCA challenge in the presence of an antioxidant and GSH precursor N-acetyl-L-cysteine (NAC) was found to alleviate the toxicity. This suggests that the basal as well as inducible GSH are important determinants of cellular resistance to LCA. Consistent with these findings, NAC cotreatment significantly reduced the oxidative stress-responsive ARE-reporter activity by LCA, indicating an antioxidative effect (Fig. 6C). BSO cotreatment, in contrast, further increased the reporter activity, consistent with a heightened oxidative stress (Fig. 6c).

    Fig. 6. Increased GSH production is a cellular protective mechanism against bile acid toxicity. a, cellular GSH levels of HepG2 after 24-h treatment with 75 µM LCA, 100 µM CDCA, or 200 µM lipoic acid (LA; positive control). BSO (60 µM), a GSH synthesis blocker, was used as negative control. *, p < 0.05, significantly different from DMSO control by one-way ANOVA followed by post hoc test. Mean and S.E.M. (n = 4). b, cell viability of HepG2 treated with toxic LCA (100 µM) together with 0.4 mM NAC, or 60 µM BSO, or with overnight pretreatment with 60 µM BSO (pre-BSO). Fluorescent units were first subtracted by those of respective treatment control (i.e., DMSO, NAC, or BSO alone), and they are expressed as percentage of change to individual baseline/pretreatment values. *, p < 0.05, significantly different from LCA treatment by one-way ANOVA followed by post hoc test. Mean ± S.E.M. of three independent experiments with four determinations. c, ARE-reporter activity in HepG2 treated with LCA with or without 0.4 mM NAC or 60 µM BSO. tert-BHP (100 µM), a peroxide radical generator, was used to show that increased cellular oxidative stress led to increased ARE reporter activity. y-axis, -fold change in the ratio of luciferase activity (relative luciferase unit) from those transfected with basic pGL3 promoter construct and treated with vehicle DMSO. *, p < 0.05, significant difference. Chemical treatments were for 8 h. Mean and S.E.M. (n = 3–4).

    The discovery of bile acids as key signaling molecules in the enterohepatic circulation system reveals a critical role of hepatic and intestinal xenobiotic nuclear receptors in the metabolism and detoxification of bile acids (Chawla et al., 2000). Particularly, the cytotoxic hydrophobic bile acids CDCA and LCA have been shown to be ligands and potent inducers of these receptors. LCA, at physiological and nontoxicological concentrations (5–30 µM), can activate FXR (Makishima et al., 1999, 2002) and VDR (Makishima et al., 2002), indicating their crucial role in physiological handling of this bile acid. The major detoxification routes of LCA, i.e., sulfation by sulfotransferase 2A and 7-hydroxylation by CYP3As, are coordinated by VDR (Makishima et al., 2002; Echchgadda et al., 2004). FXR, which induces the hepatic bile salt export pump BSEP (Ananthanarayanan et al., 2001) and down-regulates the bile-synthesizing enzyme CYP7A1 (Makishima et al., 1999), works to prevent intracellular accumulation of bile acids.

    Interestingly, at higher and toxicological concentrations of LCA (50 µM) and CDCA (100 µM), which potentially cause cell injury, PXR (Staudinger et al., 2001; Makishima et al., 2002) and Nrf2, as shown in this study, are found to be activated. The activation of PXR and Nrf2 induces the major hydroxylation enzymes CYP3As and antioxidative genes (Eloranta and Kullak-Ublick, 2005; Kensler et al., 2007), which may represent an important adaptive mechanism of cellular defense against toxic bile acids. Furthermore, we observed that induction of multiple bile salt/conjugate efflux transporters such as ATP-binding cassette (ABC) transporters ABCC2, ABCC3, and ABCG2 by bile acids is dependent on Nrf2 (K. P. Tan, G. Woodland, M. Yang, K. Kosuge, M. Yamamoto, and S. Ito, unpublished data). Hence, the collective induction of cytoprotective genes by Nrf2 and PXR seems to set off a second line of protection against possible progression of bile acid toxicity toward irreversible cell death.

    In this study, we showed for the first time that many bile acids, more potently LCA, CDCA, and DCA, are capable of activating redox-sensitive Nrf2. We also provided in vivo evidence that LCA is able to activate Nrf2, inducing similar target genes observed in in vitro studies. Because the induction of Nrf2 target genes by LCA in vivo was found to precede and sustain through biochemically and histologically overt liver injury, the collective induction of these antioxidative genes may be an integral part of cell defense against bile acid toxicity and hepatic injury. Previous studies have reported an increased intracellular production of detrimental hydroperoxides in isolated rat hepatocytes with hydrophobic bile acid exposure (Sokol et al., 1995), an observation in consensus with the increased oxidative stress byproducts in the liver of patients with cholestasis (Vendemiale et al., 2002). Because Nrf2 activation is indicative of cellular antioxidative response, our study provides molecular evidence linking mechanism of bile acid toxicity to oxidative stress.

    We further showed that induction of hepatic GCL subunits via Nrf2, which provokes GSH biosynthesis, can increase hepatocyte resistance and survival during excessive bile acid exposure. The essential role of GSH in hepatic protection against injury and oxidative xenobiotic insults has been well exemplified (Huang et al., 2001; Glosli et al., 2002). In agreement, in vivo knockout of Nrf2 enhances sensitivity of death receptor-induced hepatic apoptosis as a result of decreased GSH levels (Morito et al., 2003). GSH is also known to protect against mitochondrial injury, a major mechanism of bile acid toxicity (Palmeira and Rolo, 2004). A fraction of cytosolic GSH that becomes mitochondrial GSH is crucial in the defense of oxidant-induced mitochondrial-mediated cell death (Fernandez-Checa and Kaplowitz, 2005). In addition, Nrf2 activation has been shown to protect mitochondria by preventing inhibition of mitochondrial complex II upon exposure to oxidative neurotoxins (Calkins et al., 2005). In intestinal mucosa, cellular GSH has an essential role in maintaining epithelial integrity, transport activity, and metabolism of and susceptibility to luminal toxins (Aw, 2005). Overall, our study, coupled with supportive evidence from recent literature, suggests that protection conferred by hepatic and intestinal Nrf2 against bile acid-induced oxidative stress is, at least partly, achieved by increasing GSH levels.

    The simultaneous induction of other Nrf2 target genes may work in concert with GCL subunits to combat bile acid-induced oxidative stress and facilitate adaptive responses. Of particular importance is TRx1, an enzyme engaged in NADPH-dependent catalysis of various redox proteins (Rundlöf and Arner, 2004). It has been shown to act as a key adaptation-promoting mediator for prior exposure to 4-hydroxynonenal, a reactive lipid peroxidation-derived molecule, in inducing cellular tolerance to future oxidative stress attack (Chen et al., 2005). Indeed, intermediate cellular stress has recently been proposed to provide an adaptation advantage by invoking enhanced cellular survival/tolerance mechanisms (Schoemaker et al., 2003; Chen et al., 2005). Activation of nuclear factor B as well as Nrf2 has been shown to play an important role in this adaptation process. The drastic induction of TRx1 observed in mice upon acute exposure to toxic LCA in this study may indicate a critical role of this enzyme in adaptation process against LCA toxicity. To address whether and how this process is taking place, future studies are needed.

    The precise mechanisms by which toxic bile acids activate Nrf2 remain a subject of future studies. Enormous production of ROS from mitochondrial stress has long been accounted for the main source of oxidative stress induced by bile acids (Palmeira and Rolo, 2004). Insurgence of these ROS potentially targets the cysteine oxidative-sensors of Keap1, an actin-anchored cytosolic sequester that facilitates Nrf2 degradation by ubiquitin-proteosome pathway, which leads to liberation and activation of Nrf2 (Kensler et al., 2007). In addition, subsets of both conjugated and unconjugated bile acids have been shown to activate multiple kinase signaling pathways such as protein kinase C, extracellular signal-regulated kinase 1/2, mitogen-activated protein kinase, p38 mitogen-activated protein kinase, c-Jun NH2-terminal kinase, and/or phosphatidylinositol 3-kinase/AKT (Debruyne et al., 2002; Dent et al., 2005). These signaling pathways have been shown as well to influence the stability of Nrf2-Keap1 complex and to post-transcriptionally regulate the Nrf2 target genes (Kensler et al., 2007).

    In summary, we characterized a molecular cell defense event associated with bile acid-provoked oxidative stress. Exposure to cytotoxic bile acids in the liver and intestinal cells was shown here to cause Nrf2 activation, thereby up-regulating a battery of cytoprotective genes, particularly GCL subunits, to enhance cell survival at the emergence of oxidative stress.

    Acknowledgements

    We thank Dr. Jawed Alam for providing expression vector plasmids, Dr. Patricia Harper for providing Hepa1c1c7 cells, and Christopher Tierney (Dharmacon RNA Technologies) for technical assistance in RNA interference studies.

    ABBREVIATIONS: LCA, lithocholic acid; CDCA, chenodeoxycholic acid; FXR, farnesoid X receptor; VDR, vitamin D receptor (VDR); PXR, pregnane X receptor; ROS, reactive oxygen species; Nrf2, nuclear factor (erythroid-2 like) factor 2; ARE, antioxidant-responsive element; GCL, glutamate cysteine ligase; GCLC, glutamate cysteine ligase catalytic subunit; GCLM, glutamate cysteine ligase modulatory subunit; GSH, glutathione; MEM, minimal essential medium; FBS, fetal bovine serum; DMSO, dimethyl sulfoxide; TBL, total bilirubin; ALT, alanine aminotransferase; AST, aspartate aminotransferase; PCR, polymerase chain reaction; NQO1, NAD(P)H quinone oxidoreductase; TRx1, thioredoxin reductase 1; siRNA, small interfering RNA; siNfr2, small interfering against Nrf2; siCtr, small interfering RNA control; BHQ, tert-butylhydroquinone; NTCP, Na(+)-dependent taurocholate cotransporting polypeptide; GCA, glycocholic acid; GCDCA, glycochenodeoxycholic acid; ChIP, chromatin immunoprecipitation; bp, base pair(s); LDH, lactate dehydrogenase; MES, 2-(N-morpholino)ethanesulfonic acid; ANOVA, analysis of variance; DCA, deoxycholic acid; GST, glutathione transferase; UDCA, ursodeoxycholic acid; CA, cholic acid; BSO, buthionine sulfoximine; ABC, ATP-binding cassette; NAC, N-acetyl-L-cysteine.

【参考文献】
  Ananthanarayanan M, Balasubramanian N, Makishima M, Mangelsdorf DJ, and Suchy FJ (2001) Human bile salt export pump promoter is transactivated by the farnesoid X receptor/bile acid receptor. J Biol Chem 276: 28857–28865.[Abstract/Free Full Text]

Aw TY (2005) Intestinal glutathione: determinant of mucosal peroxide transport, metabolism, and oxidative susceptibility. Toxicol Appl Pharmacol 204: 320–328.[CrossRef][Medline]

Beresford GW and Boss JM (2001) CIITA coordinates multiple histone acetylation modifications at the HLA-DRA promoter. Nat Immunol 2: 652–657.[CrossRef][Medline]

Brown NS and Bicknell R (2001) Hypoxia and oxidative stress in breast cancer. Oxidative stress: its effects on the growth, metastatic potential and response to therapy of breast cancer. Br Cancer Res 3: 323–327.[CrossRef]

Calkins MJ, Jakel RJ, Johnson DA, Chan K, Kan YW, and Johnson JA (2005) Protection from mitochondrial complex II inhibition in vitro and in vivo by Nrf2-mediated transcription. Proc Natl Acad SciUSA 102: 244–249.[Abstract/Free Full Text]

Chan K, Han XD, and Kan YW (2001) An important function of Nrf2 in combating oxidative stress: detoxification of acetaminophen. Proc Natl Acad Sci U S A 98: 4611–4616.[Abstract/Free Full Text]

Chawla A, Saez E, and Evans RM (2000) "Don't know much bile-ology". Cell 103: 1–4.[CrossRef][Medline]

Chen ZH, Saito Y, Yoshida Y, Sekine A, Noguchi N, and Niki E (2005) 4-Hydroxynonenal induces adaptive response and enhances PC12 cell tolerance primarily through induction of thioredoxin reductase 1 via activation of Nrf2. J Biol Chem 280: 41921–41927.[Abstract/Free Full Text]

Debruyne PR, Bruyneel EA, Karaguni IM, Li X, Flatau G, Muller O, Zimber A, Gespach C, and Mareel MM (2002) Bile acids stimulate invasion and haptotaxis in human colorectal cancer cells through activation of multiple oncogenic signaling pathways. Oncogene 21: 6740–6750.[CrossRef][Medline]

Dent P, Fang Y, Gupta S, Studer E, Mitchell C, Spiegel S, and Hylemon PB (2005) Conjugated bile acids promote ERK1/2 and AKT activation via a pertussis toxin-sensitive mechanism in murine and human hepatocytes. Hepatology 42: 1291–1299.[CrossRef][Medline]

Echchgadda I, Song CS, Roy AK, and Chatterjee B (2004) Dehydroepiandrosterone sulfotransferase is a target for transcriptional induction by the vitamin D receptor. Mol Pharmacol 65: 720–729.[Abstract/Free Full Text]

Eloranta JJ and Kullak-Ublick GA (2005) Coordinate transcriptional regulation of bile acid homeostasis and drug metabolism. Arch Biochem Biophys 433: 397–412.[CrossRef][Medline]

Erickson AM, Nevarea Z, Gipp JJ, and Mulcahy RT (2002) Identification of a variant antioxidant response element in the promoter of the human glutamate-cysteine ligase modifier subunit gene. Revision of the ARE consensus sequence. J Biol Chem 277: 30730–30737.[Abstract/Free Full Text]

Fernandez-Checa JC and Kaplowitz N (2005) Hepatic mitochondrial glutathione: transport and role in disease and toxicity. Toxicol Appl Pharmacol 204: 263–273.[CrossRef][Medline]

Glosli H, Tronstad KJ, Wergedal H, Muller F, Svardal A, Aukrust P, Berge RK, and Prydz H (2002) Human TNF-alpha in transgenic mice induces differential changes in redox status and glutathione-regulating enzymes. FASEB J 16: 1450–1452.[Abstract/Free Full Text]

Hofmann AF (2004) Detoxification of lithocholic acid, a toxic bile acid: relevance to drug hepatotoxicity. Drug Metab Rev 36: 703–722.[CrossRef][Medline]

Huang ZZ, Chen C, Zeng Z, Yang H, Oh J, Chen L, and Lu SC (2001) Mechanism and significance of increased glutathione level in human hepatocellular carcinoma and liver regeneration. FASEB J 15: 19–21.[Free Full Text]

Kannan S and Jaiswal AK (2006) Low and High dose UVB regulation of transcription factor NF-E2-related factor 2. Cancer Res 66: 8421–8429.[Abstract/Free Full Text]

Kawalek JC, Hallmark RK, and Andrews AW (1983) Effect of lithocholic acid on the mutagenicity of some substituted aromatic amines. J Natl Cancer Inst 71: 293–298.[Medline]

Kensler TW, Wakabayashi N, and Biswal S (2007) Cell survival responses to environmental stress via the Keap1-Nrf2-ARE pathway. Annu Rev Pharmacol Toxicol 47: 89–116.[CrossRef][Medline]

Kozoni V, Tsioulias G, Shiff S, and Rigas B (2000) The effect of lithocholic acid on proliferation and apoptosis during the early stages of colon carcinogenesis: differential effect on apoptosis in the presence of a colon carcinogen. Carcinogenesis 21: 999–1005.[Abstract/Free Full Text]

Kr?henbühl S, Fischer S, Talos C, and Reichen J (1994) Ursodeoxycholate protects oxidative mitochondrial metabolism from bile acid toxicity: dose-response study in isolated rat liver mitochondria. Hepatology 20: 1595–1601.[Medline]

Kullak-Ublick GA, Beuers U, and Paumgartner G (1996) Molecular and functional characterization of bile acid transport in human hepatoblastoma HepG2 cells. Hepatology 23: 1053–1060.[CrossRef][Medline]

Lee JM, Li J, Johnson DA, Stein TD, Kraft AD, Calkins MJ, Jakel RJ, and Johnson JA (2005) Nrf2, a multi-organ protector? FASEB J 19: 1061–1066.[Abstract/Free Full Text]

Livak KJ and Schmittgen TD (2001) Analysis of relative gene expression data using real-time quantitative PCR and the 2(–delta delta C(T)) method. Methods 25: 402–408.[CrossRef][Medline]

Makishima M, Okamoto AY, Repa JJ, Tu H, Learned RM, Luk A, Hull MV, Lustig KD, Mangelsdorf DJ, and Shan B (1999) Identification of a nuclear receptor for bile acids. Science 284: 1362–1365.[Abstract/Free Full Text]

Makishima M, Lu TT, Xie W, Whitfield GK, Domoto H, Evans RM, Haussler MR, and Mangelsdorf DJ (2002) Vitamin D receptor as an intestinal bile acid sensor. Science 296: 1313–1316.[Abstract/Free Full Text]

Morito N, Yoh K, Itoh K, Hirayama A, Koyama A, Yamamoto M, and Takahashi S (2003) Nrf2 regulates the sensitivity of death receptor signals by affecting intracellular glutathione levels. Oncogene 22: 9275–9281.[CrossRef][Medline]

Mulcahy RT, Wartman MA, Bailey HH, and Gipp JJ (1997) Constitutive and betanaphthoflavone-induced expression of the human gamma-glutamylcysteine synthetase heavy subunit gene is regulated by a distal antioxidant response element/TRE sequence. J Biol Chem 272: 7445–7454.[Abstract/Free Full Text]

Palmeira CM and Rolo AP (2004) Mitochondrially-mediated toxicity of bile acids. Toxicology 203: 1–15.[CrossRef][Medline]

Ramos-Gomez M, Kwak M-K, Dolan PM, Itoh K, Yamamoto M, Talalay P, and Kensler TW (2001) Sensitivity to carcinogenesis is increased and chemoprotective efficacy of enzyme inducers is lost in nrf2 transcription factor-deficient mice. Proc Natl Acad SciUSA 98: 3410–3415.[Abstract/Free Full Text]

Rao CV, Hirose Y, Indranie C, and Reddy BS (2001) Modulation of experimental colon tumorigenesis by types and amounts of dietary fatty acids. Cancer Res 61: 1927–1933.[Abstract/Free Full Text]

Rundl?f AK and Arner ES (2004) Regulation of the mammalian selenoprotein thioredoxin reductase 1 in relation to cellular phenotype, growth, and signaling events. Antioxid Redox Signal 6: 41–52.[CrossRef][Medline]

Schoemaker MH, Gommans WM, Conde de al Rosa L, Homan M, Klok P, Teautwein C, van Goor H, Poelstra K, Haisma HJ, Jansen PL, et al. (2003) Resistance of rat hepatocytes against bile acid-induced apoptosis in cholestatic liver injury is due to nuclear factor-kappa B activation. J Hepatol 39: 153–161.[CrossRef][Medline]

Sokol RJ, Winklhofer-Roob BM, Devereaux MW, and McKim JM Jr (1995) Generation of hydroperoxides in isolated rat hepatocytes and hepatic mitochondria exposed to hydrophobic bile acids. Gastroenterology 109: 1249–1256.[CrossRef][Medline]

Staudinger JL, Goodwin B, Jones SA, Hawkins-Brown D, MacKenzie KI, LaTour A, Liu Y, Klaassen CD, Brown KK, Reinhard J, et al. (2001) The nuclear receptor PXR is a lithocholic acid sensor that protects against liver toxicity. Proc Natl Acad SciUSA 98: 3369–3374.[Abstract/Free Full Text]

Tietze F (1969) Enzymic method for quantitative determination of nanogram amounts of total and oxidized glutathione: applications to mammalian blood and other tissues. Anal Biochem 27: 502–522.[CrossRef][Medline]

Vendemiale G, Gratagliano I, Lupo L, Memeo V, and Altomare E (2002) Hepatic oxidative alterations in patients with extra-hepatic cholestasis. Effects of surgical drainage. J Hepatol 37: 601–605.[CrossRef][Medline]

作者单位：Division of Clinical Pharmacology and Toxicology, Department of Pediatrics (S.I.), Physiology and Experimental Medicine Program, Research Institute (K.P.T., M.Y., S.I.), Hospital for Sick Children; and Department of Pharmacology, Faculty of Medicine, University of Toronto, Ontario, Canada (K.P.T., S

【关键词】  Transcription

    Two vital enzymes of the CYP3A subfamily, CYP3A4 and CYP3A5, are differentially expressed in the human lung. However, the molecular mechanisms that regulate tissue-selective expression of the genes are poorly understood. The ability of the 5' upstream promoter region of these two genes to drive luciferase reporter activities in human lung A549 cells was dramatically different. The CYP3A5 promoter region activated luciferase gene expression by 10-fold over the promoterless construct, whereas the CYP3A4 promoter did not drive expression. Sequence comparisons of the promoters identified a 57-base pair insertion in the CYP3A4 promoter region (–71 to –127) that was absent in the CYP3A5 promoter. Deletion of the 57-bp motif from CYP3A4 or insertion into the CYP3A5 promoter, showed that this motif represses CYP3A4 expression in lung. EMSA analysis using nuclear extracts from either A549 cells or human lung tissues showed two specific protein/DNA complexes formed with the 32P-labeled CYP3A4 57-bp oligonucleotide. EMSA analyses identified two E-box motifs as the minimal specific cis-elements. Supershift assays with antibodies directed against known double- or single-E-box binding factors (TAL1, EF1, E2A, HEB, etc.) failed to identify this factor as a previously characterized trans-acting double E-box binding protein. These results demonstrated that the 5'-upstream region of CYP3A4 contains an active putative double E-box repressor motif, not present in the 5'-upstream region of the CYP3A5 gene, that attenuates CYP3A4 expression in the human lung. We believe that this is the first documented case in which a cytochrome P450 gene is actively repressed in a tissue-specific manner.

    Members of the cytochrome P4503A (CYP3A) subfamily are major contributors to the metabolism of a variety of therapeutic, xenobiotic, and endogenous compounds (Guengerich, 1999; Sheweita, 2000). The organ expressing the highest levels of P450 is the liver (Watkins, 1994), but the importance of extrahepatic expression is becoming a subject of great interest (Ding and Kaminsky, 2003). The human CYP3A subfamily is composed of four members, CYP3A4, CYP3A5, CYP3A7, and CYP3A43 (Nelson et al., 1996; Gellner et al., 2001). CYP3A4 plays the major metabolic role in adult livers (Watkins, 1994) and intestines (Kolars et al., 1994). CYP3A7 is the predominant fetal liver form (Stevens et al., 2003) and is polymorphically expressed in adults (Burk et al., 2002). CYP3A43 is the newest member of the CYP3A family (Gellner et al., 2001), but low levels of expression offer only minor contributions to total CYP3A metabolism. In comparison, CYP3A5 is expressed in the highest number of tissues (Kolars et al., 1994; Koch et al., 2002; Ding and Kaminsky, 2003), and its expression is limited by a single nucleotide polymorphism designated CYP3A5*3 (an A>G conversion within intron 3 of the CYP3A5 gene). However, the functional consequences of this polymorphism are controversial. In general, persons expressing the wild-type CYP3A5*1 allele have the highest levels of CYP3A protein in the liver, where 3A5 protein is roughly equal to that of 3A4; but persons expressing the homozygous CYP3A5*3/*3 allele have very low to undetectable levels of CYP3A5 protein (Kuehl et al., 2001). Therefore, one would expect that people who are homozygous for CYP3A5*3/*3 should turnover CYP3A substrates very poorly (Kuehl et al., 2001). However, this expectation has sometimes not been observed (Floyd et al., 2003; Westlind-Johnsson et al., 2003).

    The regulation of cytochrome P450 gene expression has received much attention in recent years, and the mechanisms responsible for CYP3A gene expression vary widely and are often complex (Guengerich, 1999; Quattrochi and Guzelian, 2001; Gibson et al., 2002; Goodwin et al., 2002b). To address the obvious importance of CYP3A in drug-drug interactions, a humanized mouse model of hepatic CYP3A4 regulation has been created (Zhang et al., 2003), and much work has been done in recent years to elucidate its regulatory mechanisms (Ding and Kaminsky, 2003; Schuetz, 2004; Xie et al., 2004). The 5'-flanking regions of the CYP3A subfamily of genes contain sequence motifs (i.e., cis-elements) that can regulate these genes in three generalized ways: 1) those that are involved in enhancing the enzyme's production [inducers: xenobiotic responsive enhancer module (XREM), pregnane X response element (PXRE), glucocorticoid receptor (GR), etc.] (Hukkanen et al., 2000, 2003; Goodwin et al., 2002b); 2) those that are involved in maintaining basal level expression (e.g., nuclear factor Y (NFY), nuclear factor B (NFB), constitutive androstane receptor (CAR), hepatocyte nuclear factor 3 (HNF3), etc.) (Iwano et al., 2001; Saito et al., 2001; Goodwin et al., 2002a; Bombail et al., 2004); and 3) those involved in turning genes off in the presence or absence of specific signals [e.g., CCAAT/enhancer-binding protein, -LIP, silencing mediator for retinoid and thyroid hormone receptors (SMRT)] (Chen and Evans, 1995; Chen and Li, 1998; Jover et al., 2002; Johnson et al., 2006). The combined actions of these and other cis-acting elements ensure that the proper amount of CYP3A protein is produced only at the precise time it is needed.

    The CYP3A enzymes expressed in human lung are important determinants of pulmonary carcinogenesis caused by metabolism of several inhaled xenobiotic compounds (Piipari et al., 2000; Mollerup et al., 2001; Yeh et al., 2003). Other lung diseases cause significant morbidity and mortality, and specific P450 enzymes are at least partially responsible for these diseases (Ding and Kaminsky, 2003; Yeh et al., 2003). These P450 enzymes probably participate in the metabolism and bioactivation of polycyclic aromatic hydrocarbons and other procarcinogens present in combustion products, tobacco smoke, and ambient particulate matter (Nelson et al., 1996; Guengerich and Shimada, 1998). CYP3A4 and CYP3A5 are active in the metabolic detoxication of benzo[a]pyrene, but they are also partly responsible for the activation of benzo[a]pyrene-7,8-diol to carcinogenic diol epoxides that are capable of covalently binding to DNA. CYP3A4 and CYP3A5 exhibit differences in expression patterns, both within tissues and among individuals (Ding and Kaminsky, 2003). Transcripts of CYP3A4 have not been found in respiratory epithelial cell lines, such as the human A549 lung cell line and the immortalized bronchial epithelial BEAS-2B cell line. Nor have transcripts been detected in human bronchoalveolar macrophages or peripheral blood lymphocytes (Willey et al., 1996; Anttila et al., 1997; Hukkanen et al., 1997; Hukkanen et al., 2000; Piipari et al., 2000). In contrast, CYP3A5 is consistently expressed in all of these cell types. Moreover, tissue-specific expression of the CYP3A genes is thought to be a major factor influencing interindividual variation in both drug response and lung disease susceptibility (Piipari et al., 2000; Lamba et al., 2002). Thus, the dominant mechanisms that regulate CYP3A expression in the human lung must be established to understand interindividual and inter-racial susceptibility to inhaled xenobiotics.

    Reagents. Plasmid/RNA isolation kits were purchased from QIAGEN (Valencia, CA). Bacterial artificial chromosomes containing human sequences for the CYP3A loci [AC005020, AC011904, and AF280107[GenBank] (Gellner et al., 2001)] were purchased from Incyte Genomics (Palo Alto, CA). Oligonucleotides used in PCR amplification of genomic sequences and EMSA analyses were purchased from Integrated DNA Technologies (Coralville, IA). The pGL3-luciferase reporter-plasmid and dual-luciferase reporter assay system were purchased from Promega (Madison, WI). SDS-polyacrylamide gel electrophoresis reagents (37.1:1 acrylamide/bis-acrylamide, ammonium persulfate, and N,N,N',N'-tetra-methyl-ethylene-diamine) were purchased from Bio-Rad Laboratories (Hercules, CA). Human lung tissue was acquired from consenting donors through the Rocky Mountain Donor Services (Salt Lake City, UT). Antibodies against known E-box binding proteins were generous gifts from Dr. Frans van Roy (SIP1) and Dr. Cornelis Murre (E2A and E47) or purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Platinum Pfx DNA polymerase, Platinum PCR Supermix High Fidelity, TOPO cloning kits, cell culture media, restriction enzymes, and all other molecular biology reagents were purchased from Invitrogen (Carlsbad, CA), BD Gentest (Woburn, MA), or Lonza Walkersville, Inc. (Walkersville, MD). Real-Time PCR reagents were purchased from Bioline (Randolph, MA) and Invitrogen. All sequence manipulations were conducted in silico using the Clone Manager Suite 7.0 (Scientific and Educational Software, Durham, NC) software package.

    Cell Culture. Human adenocarcinoma A549 cells were obtained from American Type Culture Collection (Manassas, VA). A549 cells (p98) were seeded in 75-cm2 flasks using cryopreserved aliquots (1 x 106 cells; P87–90) and maintained for 48 h with Dulbecco's modified Eagle's medium/nutrient mixture with Ham's F12 medium, supplemented with 10% fetal bovine serum. For subculturing, cells were trypsin-disassociated and reseeded at 10% confluence in fresh media. For transfection assays, cells were trypsin-disassociated and reseeded in 96-well plates at a concentration of 1.0 x 104 cells/well in 100 µl of the appropriate media. Normal Human Bronchial Epithelial (NHBE) cells were obtained from Lonza Walkersville and were grown in bronchial epithelial growth medium and cultured as above in flasks and dishes coated with collagen, fibronectin, and albumin. They were passaged no more than three times before transfection. Frozen primary hepatocytes were obtained from BD Gentest. They were thawed in a 75-cm2 flask according to the supplier's instructions and then reseeded as described above for A549 cells into 96-well plates in hepatocyte growth medium.

    Construction of Promoter-Luciferase Reporter Plasmids. Luciferase reporter constructs containing promoter regions from CYP3A5 (pGL3A5–218, pGL3A5–443, pGL3A5–647, pGL3A5–872, pGL3A5–1150, and pGL3A5–1365) and a fragment of the 5'-untranslated region of the gene (+31 base pairs) were adapted from previous work (Hukkanen et al., 2003). Chimeric CYP3A4-promoter- and CYP3A5-promoter-luciferase reporter plasmids were prepared by PCR amplification of the 5'-flanking regions of these genes from bacterial artificial chromosomes (i.e., AF_280107 for 3A4 clones; AC_005020 for 3A5 clones) using the primers listed in Tables 1 and 2. Both homology maps comparing the CYP3A4 and CYP3A5 promoters and stringent MATCH/TRANSFAC analyses were conducted to predetermine putative transcriptional binding motifs to guide primer design. Cloning primers introduced 5'-restriction sites for subsequent insertion into the multiple cloning site of the pGL3-Basic vector. For CYP3A4, a 5'-NheI site and a 3'-HindIII site were incorporated into the forward and reverse primers, respectively; for CYP3A5 constructs, a 5'-MluI site and a 3'-BglII site were used. pGL3-3A4 constructs encompassed regions surrounding the transcriptional start site (TSS) (i.e., from 3'-TSS + 118 to 5'-TSS, –35, –59, –72, –80, –150, –170, –222, –2795, and –13kb, respectively). A larger pGL3-3A5 5.3-kb plasmid was also created by PCR-amplification of a 2.3-kilobase pair region (bases 10,195 to 13,107) from a bacterial artificial chromosome containing CYP3A5 (AC_005020). Using the oligonucleotides Forward-3A5 5.3 kb-KpnI (5'-ggtaccATGCTCGTGTGCCTGATAAC-3'; sense, bases 10,195 to 10,216) and A5_P1_Reverse (5'-GCATTGCTTTGGGTAGTATGGAC-3'; antisense, bases 13,107 to 13,085), a product incorporating a KpnI restriction site at the 5'-end was created. Utilization of this new KpnI site, and a unique AflII restriction site (bases –2837 to –2842) in the CYP3A5 promoter, directed the proper insertion of this region into the original pGL3-3A5 3-kb plasmid, which was prepared separately. All plasmids were screened with restriction digestions using enzymes, which cut at least once in both the vector and the insert, and then sequenced by the University of Utah core sequencing facility.

    TABLE 1 Primer sequences used to generate the CYP3A4 nested deletion and mutation reporter plasmids

    Residues that match wild-type sequences are shown in upper case. Mutagenized residues are shown in lower case.

    TABLE 2 Primer sequences used to generate the CYP3A5 nested deletion and mutation reporter plasmids

    The mutated nucleotides are indicated in lower case. The reverse primers used were the reverse complements but were otherwise identical.

    Site-Directed Mutagenesis. Site-directed mutagenesis of the pGL3-CYP3A5–208 construct was performed with the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's instructions. The mutations were introduced using specific oligonucleotides (listed in Table 2). Correct assembly of the mutations was confirmed by sequencing.

    PCR-Based Mutations of CYP3A4 and CYP3A5 Constructs. The chimeric reporter plasmids created above were mutated to introduce or knockout the 57-bp region for the CYP3A5 and CYP3A4 constructs, respectively. Primers used in these methods are listed in Tables 1 and 2. The CYP3A5-knock-in construct was made with forward and reverse primers with 3' ends that were complementary to the boundaries of homology within the CYP3A5 promoter, but whose 5' ends were complementary to the CYP3A4 57-bp region. To reduce errors in oligonucleotide syntheses, the 57-bp insertion was divided between the forward and reverse primers, and Platinum Pfx DNA polymerase (a proof-reading polymerase that immediately terminates at the end of a template leaving blunt ends) protocols were applied as the manufacturer suggested. CYP3A4 57-bp deletion mutants were generated by long-template PCR amplification of the entire vector excluding the targeted 57-bp region. To accomplish these deletions, the wild-type CYP3A4 constructs were subjected to a deletion protocol using Platinum PCR Supermix High Fidelity containing 20 ng of original template and 20 nM concentrations of each primer (3A4–57 bp-mut F and 3A4–57 bp-mut R) in a total volume of 50 µl and subjected to an initial melting step of 94°C for 2 min, 20 cycles of amplification (94°C for 30 s; 55°C for 90 s; 68°C for 8.5 min.), and a final capping step of 68°C for 3.5 min. Ten PCR reactions were pooled and the template strands were removed by digestion with DpnI. The mutated products were purified on a 0.7% agarose gel with an expected product size of 8 kb. Because the initial PCR reaction mixture contained Taq DNA polymerase, the resulting products also contained 3'-polyadenosine overhangs that were subsequently removed with mung bean nuclease (New England Biolabs, Ipswich, MA), gel-purified as before, and then self-ligated with T4-ligase in Rapid Ligation Buffer (Promega). Due to the lack of flexibility in selection of the priming site, this protocol consistently yielded a 64-bp knockout CYP3A4 construct with 7 bp more than the 57-bp region removed (the total region that was deleted was –71 to –132), but the 64-bp knockout still closely resembled the homologous region of the CYP3A5 gene. Multiple attempts to obtain a knockout construct with precisely 57 bp deleted were not successful. Similar attempts to restore this region to these constructs using the knockin technique described above were also unsuccessful. However, a similar knockout was created using the protocol described above with 3A4-SacII (Table 1) primers to introduce two SacII restriction sites flanking the 57-bp region. Creation of the CYP3A4-Spacer construct was accomplished using the 64-bp knockout as a template, using primers that harbored the additional spacer sequence at the 5' end such that they would be included in the final PCR product. To insure that there were no other mutations introduced into the pGL3-Basic vector during these exceptionally long polymerization events, all completed promoter constructs were subcloned back into the original pGL3-Basic vector, screened with restriction enzymes, and sequenced.

    Transient Transfection and Luciferase Assay. Parallel luciferase reporter assays were conducted to compare the genetic differences between CYP3A4 and CYP3A5 transcriptional activity in cultured human lung A549 cells. Approximately 50-ml cultures of each plasmid were purified using the EndoFree Plasmid Maxi Kit (QIAGEN, Valencia, CA.), yielding 100–200 µg of DNA each. When confluence reached 70%, the A549 cells were transfected with 0.25 µg of reporter plasmid and 0.005 µg of Renilla reniformis luciferase plasmid (pRL-SV40) using FuGene 6 (Roche, Indianapolis, IN), according to the manufacturer's suggestions. Primary cell cultures (NHBE and hepatocytes) were transfected using Effectene reagent (QIAGEN), according to the manufacturer's suggestions. Cells were lysed for 36 h (A549) or 24 h (primary cultures) after the transfections, and the respective luciferase activities were determined using the dual-luciferase assay system (Promega). Firefly luciferase activities for the experimental constructs were normalized for transfection efficiency and cell loading using R. reniformis luciferase activity and total protein concentration, respectively. Data from these experiments were expressed as -fold luminescence over the activity of the promoterless pGL3-Basic reporter plasmid. The data for A549 transfections were presented as mean -fold luminescence (± S.D.) for three independent experiments performed in quadruplicate. NHBE and hepatocyte data were for a single experiment with nine replicates for each plasmid. Transfections of the mutated CYP3A5 –208 constructs to A549 cells were performed as described previously (Hukkanen et al., 2003).

    Quantitative Real-Time PCR. Total RNA was purified from 106 NHBE or A549 cells using TRIzol. One microgram of total RNA was used to synthesize first-strand cDNA using random hexomers and SuperScript II (Invitrogen), diluted 1:5 and then evaluated by quantitative real-time PCR using a Chromo-4 cycler (Bio-Rad Laboratories) and SYBR green (Invitrogen). The primers (a generous gift of Roger Gaedigk, University of Missouri, Kansas City, MO) were: for CYP3A4 detection, 5'-CTCTCATCCCAGACTTGGCCA-3' and 5'-ACAGGCTGTTGACCATCATAAAAG-3'; for CYP3A5 detection, 5'-GACCTCATCCCAAATTTGGCGG-3' and 5'-CAGGGAGTTGACCTTCATACGTT-3'; -actin was used as a housekeeping control gene (primers: 5'-GACAACGGCTCCGGCATGTGCA-3' and 5'-TGAGGATGCCTCTCTTGCTCTG-3'). We calculated the relative expression by using plasmid copy number standards.

    Electrophoretic Mobility Shift Assay. Nuclear extracts from A549 or HepG2 cells and human lung tissues were prepared as described previously (Carr et al., 2003). EMSA was performed using the gel-shift assay system from Promega essentially as described by the manufacturer. Binding reaction mixtures were preincubated at room temperature for 10 min. The mixtures contained 4 µl of nuclear extract (4 µg for lung tissue and 6 µg for cell cultures), 0.005 to 0.01 pmol 32P-labeled oligonucleotide probe, and 2 µlof5x binding buffer [50 mM Tris-HCl, pH 7.5, 250 mM NaCl, 5 mM MgCl2, 2.5 mM EDTA, 2.5 mM dithiothreitol, 20% glycerol, and 0.25 mg/ml poly(dI-dC)·poly(dI-dC)] in a total volume of 10 µl. For competition experiments, a 100-fold molar excess of unlabeled double-stranded oligonucleotide was incubated for 15 min with nuclear extract before the addition of 1 µl of 32P-labeled oligonucleotide probe (0.005–0.01 pmol). The mixtures were incubated for another 20 min at room temperature. Immediately before electrophoresis, gel loading dye (25 mM Tris-HCl, pH 7.5, 0.02% bromphenol blue, and 4% glycerol) was added to all binding reaction mixtures. The protein/DNA complexes and unbound probes were separated by nondenaturing gel electrophoresis using 4% polyacrylamide gels (1–8 V/length in centimeters) and detected by autoradiography.

    Double-stranded oligonucleotides comprising the consensus binding sequences for AP1, NFB, Sp1, OCT1, and the 2F1-LSF (Carr et al., 2003) were included (175-fold molar excess) as nonspecific competitors in the gel shift assay system (Promega). Sequence-specific competitive oligonucleotides (1–11, Table 3) were generated by mutating five base pairs at a time, using the conversion of A>C and T>G; competitors M9 to M55 (Table 3) contain single base pair mutations corresponding to the distance 5'- to 3' within the 57-bp insertion; and E-box knockout competitors were created by converting the consensus sequence (CACCTG) to a sequence that is not known to bind any transcription factor (AAAAT). Competitive oligonucleotide probes were synthesized by either Integrated DNA Technologies (Coralville, IA) or the University of Utah Core Research Facilities. Sequences of the DNA probes used in competitive EMSA experiments are listed in Table 3.

    TABLE 3 Sequences of the competitive oligonucleotides used in EMSA

    The two putative E-boxes are underlined in the wild-type sequence. Mutagenized residues are shown in bold type.

    Fig. 1. Transcriptional activity of pGL3-CYP3A4 and pGL3-CYP3A5 constructs transiently transfected into human lung A549 cells as measured using the Dual Luciferase reporter system (Roche). Expression levels were normalized to activity of the cotransfected R. reniformis luciferase vector, and are expressed as -fold of luciferase activity over the promoterless pGL3-Basic plasmid. For all experiments, pGL3-CYP3A5 constructs (A) expressed considerably higher luciferase activity than that of pGL3-CYP3A4 constructs (B) (50-fold versus 2-fold, respectively). Error bars represent S.D. of four replicate experiments run in quadruplicate.

    Transcriptional Activity of the 5' Promoter Regions of CYP3A4 and CYP3A5 in Human Lung A549 Cells. Transient transfection of the upstream promoter elements of the CYP3A4 and CYP3A5 luciferase constructs showed marked differences in the ability of these two promoters to drive luciferase expression in human lung A549 cells. On average, the pGL3-CYP3A5 promoter increased luciferase activity by more than 40-fold (Fig. 1A). In contrast, only the CYP3A4–80 construct expressed luciferase at a higher level (2.8-fold) than the promotorless control vector (Fig. 1B), which was significantly lower that the CYP3A5 construct. Comparison of the constructs with different lengths of the CYP3A5 5' flanking region showed that the majority of transcriptional activity was attributable to the proximal –208 base pairs. Sequence comparisons of the CYP3A4 and CYP3A5 5'-upstream regions revealed 85% homology between the first 500 base pairs of the two promoters. The most obvious difference between these two promoter regions is a 57-bp insertion within the promoter of CYP3A4 (–71 to –127) that contains multiple near-consensus sequences of recognized transcription factors, including a functional C/EBP element (–121 to –130) (Rodríguez-Antona et al., 2003), a putative Sp1 binding site (Bombail et al., 2004), and two putative E-box motifs (Fig. 2). A similar region is also found in the CYP3A7 promoter, except that it has a slightly smaller 56-bp insertion, and the proximal E-box is not conserved.

    Fig. 2. Schematic representation of the first 150 base pairs of the CYP3A4 and CYP3A5 5'-upstream regions. The CYP3A4 sequence has a series of subtle differences (i.e., one or two base pairs) in the core binding sequences of the CCAAT-box, BTE-box, and the ER6 motif that have been previously shown (Iwano et al., 2001) to cooperate in driving basal expression of CYP3A5 in human hepatoma HepG2 cells. The most obvious difference between these two regions is a 57-base pair insertion (–71 to –127) in the CYP3A4 promoter, which includes another intact CCAAT-box (–119 to –124) [shown to constitute a portion of the functional C/EBP element (–121 to –130) for CYP3A4 regulation in HepG2 cells (Bombail et al., 2004)] and two E-box binding motifs, which are in phase, two full turns of the DNA double helix apart. Sizes and distances of salient features are not drawn to scale.

    We have previously shown that the CYP3A5 pseudogene promoter, highly similar to CYP3A5 promoter, is transcriptionally inactive (Hukkanen et al., 2003). In the current work, we used this information to identify regions in the CYP3A5 promoter that are potentially required for the repression of transcription in A549 cells. Comparison of the promoter sequence present in the pGL3-CYP3A5 –208 construct with the corresponding CYP3A5 pseudogene sequence revealed 21 different nucleotides preceding the TATA box (Fig. 3a). All of these nucleotides in the pGL3-CYP3A5 –208 construct were mutated into the corresponding nucleotide in the pseudogene individually or in combination of two or more closely located nucleotides. The mutated constructs were transfected into A549 cells, and the luciferase activities were compared with the wild-type construct. All mutations except 3 and 20,21 significantly decreased luciferase activity (p < 0.05, one-way ANOVA, LSD post hoc test). Mutations 2, 15,16, and 18,19 decreased transcriptional activity more than 50%, and mutation 17 decreased transcriptional activity 84%. These data suggest that the differences in the relative transcriptional activities of the CYP3A4 and CYP3A5 promoters (2- versus 40-fold over the control vector, respectively) are at least partially attributable to the low promoter activity of the CYP3A4 cis-elements. Two of these elements are the previously published CCAAT-box and BTE that were crucial to the basal expression of the CYP3A5 gene in HepG2 cell lines (Iwano et al., 2001). However, these data also confirmed that the complete 57-bp insertion is required for the full repression of the CYP3A4 gene in A549 lung cells, because although a single nucleotide mutation (mutation 17, Fig. 3) in the CCAAT-box of the CYP3A5 promoter did decrease transcription, the remaining level of expression was still higher than that of the CYP3A4 promoter.

    Fig. 3. Effect of site-directed mutations on CYP3A5 promoter transcriptional activity. A, the proximal –208 bp of the CYP3A5 promoter (H3A5) was compared with the corresponding region of the CYP3A5 pseudogene (H3A5P1). The differing nucleotides are numbered, and the transcription factor binding sites known to be important for CYP3A5 transcriptional activation in liver (Iwano et al., 2001) are indicated. B, mutated pGL3-CYP3A5 –208 constructs were transfected into A549 lung cells and the luciferase activities were compared with the luciferase activity produced by the wild-type (CYP3A5–208) construct. Luciferase activities were normalized to the activity of the cotransfected R. reniformis luciferase vector, and are expressed as -fold luciferase activity over the wild type. The data represent means ± S.D. of four separate replicates. The experiment was repeated three times with similar results.

    Function of the CYP3A4 57-bp Insertion in the Repression of CYP3A4 Gene Expression. We examined the repressive ability of the CYP3A4 57-bp insertion in the human A549 cell line by creating a series of mutated CYP3A4 and CYP3A5 promoter luciferase constructs. Introduction of the CYP3A4 57-bp insertion into the analogous position of a CYP3A5 promoter construct (pGL3–3A5–3000) significantly reduced luciferase-driven expression of that construct in human A549 lung cells by roughly 50% (Fig. 4). Similar activity changes were observed for mutations in the smaller pGL3-3A5–433 and pGL3–3A5–862 luciferase constructs (Fig. 4). These data demonstrated that the CYP3A4 57-bp insertion has direct functional consequences on the low expression of CYP3A4 in human lung cells.

    Fig. 4. Insertion of the CYP3A4 57-bp insertion into the concomitant region of the CYP3A5 promoter reduces expression in the human lung adenocarcinoma A549 cell line. Salient features of the CYP3A5/luciferase reporter constructs are depicted to the left of the bar graph with individual numbers representing relative distance from the published transcriptional start site for CYP3A5. Bar graphs represent the relative -fold activity over the promoterless pGL3-Basic vector. Data represent means ± S.D. from quadruplicate experiments that have been normalized for transfection efficiency and protein content.

    In comparison, deletion of various putative cis-elements within the 57-bp region of the CYP3A4 promoter constructs altered expression in a more complex manner (Fig. 5). For example, when the 57-bp region had been completely excised from the CYP3A4 promoter (i.e., the 64-bp knockout construct), no increase in luciferase activity was observed compared with wild type (p > 0.1; unpaired t test with equal variance). However, if the C/EBP binding motif (–121 to –130), shown to be essential for trans-activation in HepG2 liver cells (Rodríguez-Antona et al., 2003), was maintained, but the rest of the motif deleted (i.e., the SacII construct; Fig. 5), luciferase expression increased substantially (p < 0.05; unpaired t test with equal variance). Moreover, a similar CYP3A4-promoter construct harboring specific mutations that destroyed the two E-boxes, but maintained the C/EBP motif, also increased expression (data not shown). The CYP3A4-Spacer construct was made to evaluate the hypothesis that changes in reporter gene activity by deletion of the 57-bp region were not caused by alterations of simple spatial interactions of cis-elements through the shortening of this region of the CYP3A4 promoter. When this construct was tested, no difference in luciferase expression was observed (p > 0.1; unpaired t test with equal variance; Fig. 5). Together, these findings suggest that this region is functionally active in A549 cells and is therefore at least partially responsible for the repression of CYP3A4 expression within the human lung. The functionality of this region was validated by transfection experiments in primary human lung and liver cell cultures. (Fig. 6). Deletion of the region significantly (p < 0.05, one-way ANOVA, LSD post hoc test) increased expression of a CYP3A4 promoter construct in NHBE cells but not in primary hepatocytes. Likewise, NHBE expression of a CYP3A5 promoter construct was significantly (p < 0.05, one-way ANOVA, LSD post hoc test) decreased by insertion of this promoter element, whereas the insertion did not change hepatocyte expression. Quantitative real-time PCR was used to confirm the intrinsic expression of CYP3A5 in A549 cells (9000 copies/µg of total RNA) and NHBE cells (850 copies/µg of total RNA), whereas CYP3A4 was not detected in either the tumor cells or the primary lung cells (Fig. 7).

    Fig. 5. Transcriptional activity of pGL3–3A4 constructs, with and without the CYP3A4 57-bp region, transiently transfected into human lung A549 cells as measured using the Dual Luciferase reporter system (Roche). When the 57-bp region had been completely excised from the CYP3A4 promoter, no increase in luciferase activity was observed (64-bp Knockout and Spacer; p > 0.1). However, if the C/EBP binding motif (–121 to –130), shown to be essential for trans-activation in HepG2 liver cells (Rodríguez-Antona et al., 2003) was maintained (SacII construct), expression levels were increased by approximately 4-fold compared with the complete CYP3A4 construct (p < 0.05 significantly different compared with wild-type pGL3–3A4 expression levels). Expression levels were normalized to activity of the cotransfected R. reniformis luciferase vector and are expressed as -fold luciferase activity divided by the promoterless pGL3-Basic plasmid. Error bars represent the S.D. of four replicates that were normalized for transfection efficiency and protein content. The experiment was repeated twice with similar results.

    Fig. 6. Transcriptional activity of wild-type pGL3–3A4 and pGL3–3A5 constructs, with and without the CYP3A4 57-bp region, transiently transfected into primary cell cultures. The largest (13 kb) CYP3A4 promoter (a kind gift from Dr. Tetsuya Kamataki, Sapporo, Japan), which presumably contained all viable transcriptional elements, did not enhance luciferase expression in NHBE cells. The SacII deletion construct (57-bp knockout) significantly (p < 0.05) increased luciferase activity over the intact CYP3A4 construct in NHBE cells, but this deletion did not alter promotion in primary hepatocytes, both of which showed activity above that of the control vector, pGL-3 Basic. Insertion of the CYP3A4 57-bp region (57 bp inserted) into the wild-type 3-kb pGL3–3A5 construct significantly decreased (p < 0.05) luciferase activity in transfected NHBE cells but not in primary hepatocytes. Data represent means ± S.D. from nine replicates that were normalized for transfection efficiency.

    Fig. 7. Quantitative real-time PCR analysis of CYP3A4 and CYP3A5 transcripts in primary and immortalized human lung cells. Total RNA was extracted from 106 NHBE or A549 cells, and 1 µg of total RNA was used to synthesize first-strand cDNA and then analyzed by quantitative real-time PCR. Copy numbers were determined by extrapolation from a standard curve that was constructed with known amounts of CYP3A4 or CYP3A5 plasmid DNA, and standardized to actin levels. N.D., not detected (the mRNA amounts for CYP3A4 were below detectable levels in both lung cell types).

    Analysis of Factor Binding to the CYP3A4 57-bp Region by EMSA. To demonstrate the specificity of nuclear factor(s) binding to the putative cis-element(s) identified by the luciferase experiments, EMSA experiments were conducted using 32P-labeled double-stranded oligonucleotides corresponding to the wild-type sequence for the CYP3A4 57-bp insertion (–71 to –127), shown in Table 3, and nuclear extracts from A549 lung cells and whole-lung nuclear extracts. When radiolabeled wild-type CYP3A4 57-bp probe was incubated with nuclear extract from either source, two sequence-specific DNA-protein complexes were observed (Fig. 8; data from whole-lung extracts not shown). Binding was inhibited by the addition of a 100-fold molar excess of unlabeled wild-type oligonucleotide. In contrast, complex formation was not inhibited by the addition of 175-fold molar excess of unlabeled competitors harboring consensus binding sequences for AP1, NFB, Sp1, OCT1, and the CYP2F1-LSF1 motif (Table 3, Fig. 8). These results demonstrate the specific binding of nuclear factors within the human lung and human lung adenocarcinoma A549 cells to the CYP3A4 57-bp insertion. Furthermore, because consensus oligonucleotides harbor the highest affinity binding sites for their respective nuclear factors, these data also strongly suggest that the trans-acting element(s) is not AP1, NFB, Sp1, OCT1, or CYP2F1-LSF1.

    Fig. 8. Competitive EMSA assays of serial mutations of the CYP3A4 57-bp promoter element. Putative E-boxes within the CYP3A4 57-bp insertion are capable of binding in a sequence-specific manner with human lung adenocarcinoma A549 cell nuclear proteins. The 32P-labeled CYP3A4 57-bp probe was incubated with nuclear extracts (6 µg of total protein) prepared from A549 lung cells with various competitors. When competitors were included in the incubations, they were added 10 min before the addition of the 32P-labeled CYP3A4 57-bp probe. Therefore, the ability to compete for, rather than displace, the 32P-labeled CYP3A4 57-bp probe for A549 nuclear proteins was assayed. The two major protein/DNA complexes are marked with arrows. Competitive oligonucleotides were created by mutating five consecutive base pairs at a time and are labeled with numbers above the lanes that correspond to the appropriate mutated sequences shown in Table 3. The CCAAT-box and the two putative E-box motifs are labeled above the 57-bp sequence, and the consensus sequences are underlined. The five base pairs that were mutated for each competitive incubation are identified above each lane and demarcated with vertical lines. Notations are as follows: probe, 32P-labeled CYP3A4 57-bp probe only; A549, 32P-labeled CYP3A4 57-bp probe incubated with A549 nuclear extracts; Cold, 100-fold molar excess of unlabeled CYP3A4 57 bp; and NFB, 175-fold molar excess of NFB consensus oligonucleotide.

    It seems feasible that selective expression of the CYP3A5 gene in lung cells could be controlled by binding of specific transcription factors in lungs that are different from protein factors of liver cells. Therefore, we isolated nuclear proteins from the human liver cell line, HepG2, and compared binding of these proteins, and the nuclear factors from A549 cells, to the labeled 57-bp probe. The EMSA protein/DNA bands with liver nuclear factors (Fig. 9, lanes 4–6) elicited considerably less binding than A549 extracts (Fig. 9, lanes 1–3), and the two protein/DNA bands that were formed had considerably faster mobilities than the two bands using lung proteins. Thus, although liver transcription proteins seemed to bind specifically (the binding was abolished by 100-fold excess unlabeled probe) to the 57-bp motif, the proteins that bound seemed to have lower molecular weights than lung cell transcription factors.

    Fig. 9. EMSA analysis of lung and liver cellular nuclear extract binding to the 57-bp motif. Nuclear extracts (6 µg) from A549 (lanes 1–3) or HepG2 (lanes 4–6) cells were mixed with 0.005 to 0.01 pmol of 32P-labeled 57-bp oligonucleotide probe, and the protein/DNA complexes and unbound probes were separated by nondenaturing gel electrophoresis using 4% polyacrylamide gels and detected by autoradiography as described under Materials and Methods. Lanes 2 and 5 are samples from incubations that contained a 100-fold excess of unlabeled specific 57-bp oligonucleotide, incubated for 15 min with nuclear extract before the addition of labeled probe. Lanes 3 and 6 are samples from incubations that contained a 100-fold excess of nonspecific oligonucleotide, incubated for 15 min with nuclear extract before the addition of labeled probe. Right-facing arrows point to the two major protein/DNA complexes of lung cell proteins with the 57-bp probe. Left-facing arrows point to the two major protein/DNA complexes of liver cell proteins with the 57-bp probe. Binding of liver nuclear extracts (lanes 4–6) to the probe was considerably weaker than lung extracts, because most of the radioactive probe stayed with the dye front at the bottom of the gel.

    To identify the essential bases for binding of A549 nuclear proteins, competitive EMSA experiments were conducted. A series of 11 mutated 57-bp oligonucleotides, designed to mutate 5 bp at a time, were incubated with the labeled wild-type 57-bp probe. These experiments (Table 3, Fig. 8) showed that only oligonucleotide probes 3, 4, and 11 were unable to block binding of transcription factors to the native 57-bp sequence. A partial loss of competitive binding was observed for oligonucleotides 6, 9, and 10, but the DNA/protein bands were much lighter than the bands corresponding to the 5-bp mutated oligonucleotides from the E-boxes. These results suggested that the two E-box motifs within this region are the putative binding regions for A549 nuclear transcriptional proteins. When oligonucleotide probes with only single mutations in either the first (Table 3, probe M15) or second (Table 3, M49 or M52) E-box elements were included in competitive EMSA assays, the single mutations in either E-box were sufficient to ameliorate competition of the transcription protein to the DNA of the 57-bp probe. These results demonstrated a requirement for the fidelity of both E-box motifs. This is a characteristic feature of double-E-box binding proteins, such as the chicken EF1 (Remacle et al., 1999). These two E-box regions are in phase, approximately two complete turns of the double helix apart. The nuclear factor(s) prepared from human lung cells bound to the two E-box motifs in a nucleotide sequence-specific manner, and it is therefore possible that a single protein, like a member of the vertebrate polycomb proteins, could simultaneously bind to both E-boxes within this region.

    In an attempt to identify the trans-acting factor involved in the active repression of the CYP3A4 gene, supershift EMSA experiments were conducted using antibodies generated against the known double-E-box binding factors (SIP1, HEB, and EF1), as well as single E-box binding proteins (e.g., E2A, Myc, MyoD, and E47). None of the antibodies retarded the mobility of the protein/DNA complex (data not shown), indicating that these are not the trans-acting factors in question. Multiple attempts to purify and identify the lung nuclear factor from lung cells or tissues by DNA affinity chromatography were not successful. When combined, these data suggest that an uncharacterized human lung specific transcription factor(s) binds specifically to a double E-box motif within the CYP3A4 57-bp insertion and actively represses CYP3A4 expression in the human lung.

    Comparisons of the promoter regions of CYP3A4 and CYP3A5 show that they share more than 90% homology within the first 1 kb of upstream regulatory sequence and yet show striking differences in expression among individuals, races, developmental stages, tissues, and cell types. The regulation of the CYP3A cassette is also very complex with regard to its response to changes in physiological conditions.

    Most studies on the CYP3A subfamily thus far have concentrated on the expression of these isoforms in the liver, where polymorphisms have been shown to be important. An intriguing ramification of previous studies (Kuehl et al., 2001) is that 75% of a population (i.e., those homozygous for CYP3A5*3) would express an extremely limited amount of CYP3A5 within their extrahepatic tissues, thus increasing the metabolic burden upon the CYP3A4 enzyme. Because CYP3A5 is probably the major CYP3A isoform expressed within human lung tissue (Anttila et al., 1997; Raunio et al., 1999), the overall metabolism of 3A substrates in lung cells would be drastically reduced or otherwise altered toward alternative metabolic pathways. Although procarcinogens metabolized by the CYP3A enzymes would not become bioactivated, drugs designed to target the lung might exhibit greater adverse side effects (e.g., long-term use of inhaled glucocorticoids might lead to superinfections) and carcinogens/toxins would exhibit extended half-lives. CYP3A individual genotypes have been correlated to lower incidence of lung and other types of cancer (Yeh et al., 2003; Keshava et al., 2004).

    As expected in A549 cells, CYP3A5-Luc constructs exhibited more than 20-fold higher luciferase activity than CYP3A4-Luc constructs. This dramatic difference in basal expression among these constructs becomes quite apparent within nested deletions containing only the first 200 base pairs of the 5'-flanking region of these two genes (Fig. 1). Luciferase constructs that contained the 3A4 promoter gradually increased in activity as nested-deletions reached –80 but then dropped dramatically thereafter. In contrast, CYP3A5 constructs rapidly reached maximal levels with the first –208 base pairs of the promoter; a pattern of expression that closely matches previous studies in HepG2 liver cells (Iwano et al., 2001). The experiments with mutated CYP3A5 promoter constructs showed that the basic transcriptional element (BTE) and especially the CCAAT box are important for CYP3A transcriptional activation in A549 cells (Fig. 4), a finding that is surprisingly similar to the mechanisms of CYP3A transcriptional regulation in HepG2 cells.

    We can conjecture about the genetic regulatory mechanisms in A549 lung cells by comparing the mechanisms governing expression of CYP3A4 and CYP3A5 in HepG2 cells. For instance, the basal regulation of CYP3A5 is governed primarily by the cooperative effects of NF-Y and specificity protein (Sp) family members binding to a CCAAT-box (–68 to –78) and the BTE (–46 to –67) in the proximal promoter of the CYP3A5 gene (Iwano et al., 2001). Analogous motifs are identifiable within the proximal promoter of the CYP3A4 gene. However, compared with the CYP3A5 promoter, the CYP3A4 CCAAT-box (–62 to –66) has a single mutation in the core binding motif, and the BTE (–36 to –57) also shows a difference in two consecutive nucleotides. Both perturbations confer markedly less expression activity than their respective CYP3A5 counterparts when tested in HepG2 cell lines (data not shown), and this is also the case in A549 cells (Fig. 1). In contrast, the transcriptional regulation of the CYP3A4 gene in HepG2 cell lines is controlled by a complex circuit of transcription factors that bind to motifs more distal to the related region of CYP3A5. Instead of being governed by the aforementioned regions, the basal expression of the CYP3A4 gene in HepG2 cell lines is only partially controlled by the binding of C/EBP and HNF-3 to a proximal ER-6 motif (–152 to –169) (Rodríguez-Antona et al., 2003); expression depends rather upon HNF-1, HNF-4, AP-1, and USF1 binding to a distal region (–10.9 to –11.4 kb), which was referred to as the "constitutive liver enhancer module of CYP3A4 (CLEM4)" (Matsumura et al., 2004). However, this is not the case in transient transfections of these same constructs in A549 cells, because the inclusion of the CLEM4 element did not increase expression (data not shown). What is important about these findings is that the underlying mechanisms responsible for the basal expression of these two genes are encoded within the sequences of the CYP3A4 and CYP3A5 promoters and that, although similar, the expression patterns within A549 lung cells are distinct from that of HepG2 liver cells.

    In a comparison of the reporter activities of CYP3A4-Luc and CYP3A5-Luc in the human lung A549 cell line, dramatic differences in promoter-driven expression were observed within the first –150 and –208 base pairs for CYP3A4 and CYP3A5, respectively. The most obvious difference between these two promoter regions is a 57-base pair insertion in the CYP3A4 promoter (–71 to –127; Fig. 2). More importantly, when introduced into the same region of the CYP3A5 promoter, this insertion reduced the expression of CYP3A5-promoter constructs by roughly 50% (Fig. 4). It is important to note that the insertion of this 57-bp region disrupts the regulatory CCAAT-box that is vital to drive basal expression in HepG2 cells (Iwano et al., 2001) and in A549s (this study). However, this insertion also includes a CCAAT-box at the 5' end that evolutionarily conserves this motif almost perfectly, suggesting either that another protein is interacting with this introduced region or that distance from neighboring motifs is fundamental to maintaining this expression mechanism.

    Two conclusions can be drawn from these observations: 1) differences between the proximal cis-elements of the CYP3A4 and CYP3A5 promoters reduce the expression of the CYP3A4 gene in human lung cells; and 2) control of the lung-specific differences in expression observed in previous studies (Kivistö et al., 1996; Anttila et al., 1997; Hukkanen et al., 2000, 2002) is a combination of these differences and the insertion of a repressor motif in the form of a 57-bp region containing two E-box motifs directly between the positive cis-elements and the transcriptional initiation site of the CYP3A4 gene. A schematic diagram illustrating the mechanism by which the CYP3A4 gene could be repressed through this region is depicted in Fig. 10A.

    Fig. 10. A, schematic representation of the LSF2 mechanism of repressing CYP3A4 in the human lung. Actions of both the proximal (Rodríguez-Antona et al., 2003) and the distal (Matsumura et al., 2004) elements converge upon an ER-6 motif and a CCAAT-box to recruit the RNA Pol II transcriptional machinery to the CYP3A4 transcriptional start site. By binding to the 57-bp insertion located immediately downstream of this region (–71 to –128), LSF2 disrupts these events through one or a combination of actions: 1) altering the recognition of the CCAAT-box by these factors, 2) physically inhibiting the assembly of the complex, or 3) aiding the condensation of chromatin to mask the cis-elements. B, the sequence highlighting the protein/DNA interactions of the CYP3A4 promoter region. Highlighted are the 57-bp insertion (red); the HNF-3, ER-6, C/EBP, and TATA transcription factor binding sites (gray); the E-box sites (underlined); and the core nucleotides for binding of A549 nuclear extracts to the 57-bp probe from EMSA experiments (Fig. 8) or the core nucleotides for binding of HepG2 nuclear proteins (yellow). The transcription start site (+1) and various cloning sites are identified as well.

    The work described herein identifies a repressor motif (–71 to –127) within the 5'-flanking region of this gene that can be bound by at least one currently unidentified nuclear protein expressed in human lung adenocarcinoma A549 cells and in normal human lung tissues. Nuclear proteins from both A549 lung cells and human lung tissues specifically bind the E-box motifs within this 57-bp region, and the fidelity of both motifs are required for the assembly of these specific regulatory protein complexes (Table 3, Fig. 8), suggesting that this region could associate with a double E-box binding protein resembling -EF1 (Remacle et al., 1999), a vertebrate polycomb protein involved in differentiation, and long-term gene silencing. EMSA supershift experiments using antibodies generated against all known members of the -EF1 family of transcription factors failed to identify the protein associating with this region (data not shown); suggesting the possibility of a novel member of the -EF1 family of transcription factors is binding to this region. Our repeated attempts to purify the protein(s) by DNA affinity chromatography were not successful, so we must refer to the protein(s) as a putative double E-box transcription factor.

    The codependence on both E-box motifs could also be associated with cooperative binding of multiple transcription factors to this region. Although the detailed interactions between these motifs have not been rigorously established, these data strongly suggest that the active repression of CYP3A4, and not the lack of a transcriptional activator, is the operative mechanism in the human lung controlling the differential expression of these two very important genes. We believe that this is the first case in which a cytochrome P4503A gene is actively "silenced" in a tissue-specific manner. We have termed this trans-repressor protein "lung-specific factor-2" (LSF2). Although the precise molecular mechanisms involved in LSF2 repression of CYP3A4 in the human lung remain unclear, we hypothesize that LSF2 actively represses the expression of the CYP3A4 gene in human lung cells.

    The underlying mechanism could exhibit its effects through three different mechanisms (Johnson, 1995): 1) interfering directly with other cis-acting elements through competition for their cognitive cis-activation domain (Sekido et al., 1997), 2) physically blocking the guided assembly (or initiation) of the transcription machinery (Coumoul et al., 2002), or 3) having some chromatin remodeling capacity (Ringrose and Paro, 2004). A number of recent observations, both genetic and biochemical, suggest a different mechanism of P450 repression might be active in HepG2 liver cells. The PXR-SMRT mechanism of CYP3A4 repression in the absence of ligand (Johnson et al., 2006), involves both sequestration of trans-acting elements and direct recruitment of histone deacetylases involved in chromatin condensation. Supershift EMSA experiments using polyclonal goat antibodies raised against SMRT did not change the mobility of the A549 nuclear protein/DNA band. Therefore, it is highly unlikely that SMRT is the trans-element involved in the pulmonary repression mechanism.

    Given that the 57-bp insertion is situated on the 3' end of the promoter, the simplest mechanism would be that LSF2 would hinder the formation of the initiation complex either through steric hindrance or through the recruitment of chromatin remodeling complexes. Active repression by LSF2 is an attractive mechanism because it can override a number of different signal cascades with the same protein, regardless of which additional transcription factors might be actively transcribed within the cell.

    Acknowledgements

    We thank Cassandra Deering for her efforts in creating some of the luciferase-reporter constructs within these studies, Dr. Ronald Hines and Dr. Brian Carr for assistance with cell culture and luciferase assay techniques, and Dr. Tetsuya Kamataki (Hokkaido University, Sapporo, Japan) for providing the wild-type pGL3-CYP3A4-13 kb luciferase-reporter construct.

    ABBREVIATIONS: P450, cytochrome P450; PCR, polymerase chain reaction; EMSA, electrophoretic mobility shift assay; kb, kilobase pair(s); bp, base pair(s); NHBE, normal human bronchial epithelial cells; AP1, activator protein-1; NFB, nuclear factor B; C/EBP, CCAAT/enhancer-binding protein; ANOVA, analysis of variance; LSD, least significant difference; BTE, basic transcriptional element; NF-Y, nuclear factor Y; Sp1, specificity protein 1; LSF, lung-specific factor; A549, human lung adenocarcinoma A549 cells.

【参考文献】
  Anttila S, Hukkanen J, Hakkola J, Stjernvall T, Beaune P, Edwards RJ, Boobis AR, Pelkonen O, and Raunio H (1997) Expression and localization of CYP3A4 and CYP3A5 in human lung. Am J Respir Cell Mol Biol 16: 242–249.[Abstract]

Bombail V, Taylor K, Gibson GG, and Plant N (2004) Role of Sp1, C/EBPalpha, HNF3, and PXR in the basal- and xenobiotic-mediated regulation of the CYP3A4 gene. Drug Metab Dispos 32: 525–535.[Abstract/Free Full Text]

Burk O, Tegude H, Koch I, Hustert E, Wolbold R, Glaeser H, Klein K, Fromm MF, Nuessler AK, Neuhaus P, et al. (2002) Molecular mechanisms of polymorphic CYP3A7 expression in adult human liver and intestine. J Biol Chem 277: 24280–24288.[Abstract/Free Full Text]

Carr BA, Wan J, Hines RN, and Yost GS (2003) Characterization of the human lung CYP2F1 gene and identification of a novel lung-specific binding motif. J Biol Chem 278: 15473–15483.[Abstract/Free Full Text]

Chen JD and Evans RM (1995) A transcriptional co-repressor that interacts with nuclear hormone receptors. Nature 377: 454–457.[CrossRef][Medline]

Chen JD and Li H (1998) Coactivation and corepression in transcriptional regulation by steroid/nuclear hormone receptors. Crit Rev Eukaryot Gene Expr 8: 169–190.[Medline]

Coumoul X, Diry M, and Barouki R (2002) PXR-dependent induction of human CYP3A4 gene expression by organochlorine pesticides. Biochem Pharmacol 64: 1513–1519.[CrossRef][Medline]

Ding X and Kaminsky LS (2003) Human extrahepatic cytochromes P450: function in xenobiotic metabolism and tissue-selective chemical toxicity in the respiratory and gastrointestinal tracts. Annu Rev Pharmacol Toxicol 43: 149–173.[CrossRef][Medline]

Floyd MD, Gervasini G, Masica AL, Mayo G, George AL, Jr., Bhat K, Kim RB, and Wilkinson GR (2003) Genotype-phenotype associations for common CYP3A4 and CYP3A5 variants in the basal and induced metabolism of midazolam in European- and African-American men and women. Pharmacogenetics 13: 595–606.[CrossRef][Medline]

Gellner K, Eiselt R, Hustert E, Arnold H, Koch I, Haberl M, Deglmann CJ, Burk O, Buntefuss D, Escher S, et al. (2001) Genomic organization of the human CYP3A locus: identification of a new, inducible CYP3A gene. Pharmacogenetics 11: 111–121.[CrossRef][Medline]

Gibson GG, Plant NJ, Swales KE, Ayrton A, and El-Sankary W (2002) Receptor-dependent transcriptional activation of cytochrome P4503A genes: induction mechanisms, species differences and interindividual variation in man. Xenobiotica 32: 165–206.[CrossRef][Medline]

Goodwin B, Hodgson E, D'Costa DJ, Robertson GR, and Liddle C (2002a) Transcriptional regulation of the human CYP3A4 gene by the constitutive androstane receptor. Mol Pharmacol 62: 359–365.[Abstract/Free Full Text]

Goodwin B, Redinbo MR, and Kliewer SA (2002b) Regulation of cyp3a gene transcription by the pregnane x receptor. Annu Rev Pharmacol Toxicol 42: 1–23.[CrossRef][Medline]

Guengerich FP (1999) Cytochrome P-450 3A4: regulation and role in drug metabolism. Annu Rev Pharmacol Toxicol 39: 1–17.[CrossRef][Medline]

Guengerich FP and Shimada T (1998) Activation of procarcinogens by human cytochrome P450 enzymes. Mutat Res 400: 201–213.[Medline]

Hukkanen J, Hakkola J, Anttila S, Piipari R, Karjalainen A, Pelkonen O, and Raunio H (1997) Detection of mRNA encoding xenobiotic-metabolizing cytochrome P450s in human bronchoalveolar macrophages and peripheral blood lymphocytes. Mol Carcinog 20: 224–230.[CrossRef][Medline]

Hukkanen J, Lassila A, Paivarinta K, Valanne S, Sarpo S, Hakkola J, Pelkonen O, and Raunio H (2000) Induction and regulation of xenobiotic-metabolizing cytochrome P450s in the human A549 lung adenocarcinoma cell line. Am J Respir Cell Mol Biol 22: 360–366.[Abstract/Free Full Text]

Hukkanen J, Pelkonen O, Hakkola J, and Raunio H (2002) Expression and regulation of xenobiotic-metabolizing cytochrome P450 (CYP) enzymes in human lung. Crit Rev Toxicol 32: 391–411.[CrossRef][Medline]

Hukkanen J, Vaisanen T, Lassila A, Piipari R, Anttila S, Pelkonen O, Raunio H, and Hakkola J (2003) Regulation of CYP3A5 by glucocorticoids and cigarette smoke in human lung-derived cells. J Pharmacol Exp Ther 304: 745–752.[Abstract/Free Full Text]

Iwano S, Saito T, Takahashi Y, Fujita K, and Kamataki T (2001) Cooperative regulation of CYP3A5 gene transcription by NF-Y and Sp family members. Biochem Biophys Res Commun 286: 55–60.[CrossRef][Medline]

Johnson AD (1995) The price of repression. Cell 81: 655–658.[CrossRef][Medline]

Johnson DR, Li C-W, Chen L-Y, Ghosh JC, and Chen JD (2006) Regulation and binding of pregnane X receptor by nuclear receptor corepressor silencing mediator of retinoid and thyroid hormone receptors (SMRT) 10. Mol Pharmacol 69: 99–108.[Abstract/Free Full Text]

Jover R, Bort R, Gomez-Lechon MJ, and Castell JV (2002) Down-regulation of human CYP3A4 by the inflammatory signal interleukin-6: molecular mechanism and transcription factors involved. FASEB J 16: 1799–1801.[Abstract/Free Full Text]

Keshava C, McCanlies EC, and Weston A (2004) CYP3A4 polymorphisms—potential risk factors for breast and prostate cancer: a HuGE review. Am J Epidemiol 160: 825–841.[Abstract/Free Full Text]

Kivist? KT, Griese EU, Fritz P, Linder A, Hakkola J, Raunio H, Beaune P, and Kroemer HK (1996) Expression of cytochrome P 450 3A enzymes in human lung: a combined RT-PCR and immunohistochemical analysis of normal tissue and lung tumours. Naunyn Schmiedebergs Arch Pharmacol 353: 207–212.[Medline]

Koch I, Weil R, Wolbold R, Brockmoller J, Hustert E, Burk O, Nuessler A, Neuhaus P, Eichelbaum M, Zanger U, et al. (2002) Interindividual variability and tissue-specificity in the expression of cytochrome P450 3A mRNA. Drug Metab Dispos 30: 1108–1114.[Abstract/Free Full Text]

Kolars JC, Lown KS, Schmiedlin-Ren P, Ghosh M, Fang C, Wrighton SA, Merion RM, and Watkins PB (1994) CYP3A gene expression in human gut epithelium. Pharmacogenetics 4: 247–259.[Medline]

Kuehl P, Zhang J, Lin Y, Lamba J, Assem M, Schuetz J, Watkins PB, Daly A, Wrighton SA, Hall SD, et al. (2001) Sequence diversity in CYP3A promoters and characterization of the genetic basis of polymorphic CYP3A5 expression. Nat Genet 27: 383–391.[CrossRef][Medline]

Lamba JK, Lin YS, Schuetz EG, and Thummel KE (2002) Genetic contribution to variable human CYP3A-mediated metabolism. Adv Drug Deliv Rev 54: 1271–1294.[CrossRef][Medline]

Matsumura K, Saito T, Takahashi Y, Ozeki T, Kiyotani K, Fujieda M, Yamazaki H, Kunitoh H, and Kamataki T (2004) Identification of a novel polymorphic enhancer of the human CYP3A4 gene. Mol Pharmacol 65: 326–334.[Abstract/Free Full Text]

Mollerup S, Ovrebo S, and Haugen A (2001) Lung carcinogenesis: resveratrol modulates the expression of genes involved in the metabolism of PAH in human bronchial epithelial cells. Int J Cancer 92: 18–25.[CrossRef][Medline]

Nelson DR, Koymans L, Kamataki T, Stegeman JJ, Feyereisen R, Waxman DJ, Waterman MR, Gotoh O, Coon MJ, Estabrook RW, et al. (1996) P450 superfamily: update on new sequences, gene mapping, accession numbers and nomenclature. Pharmacogenetics 6: 1–42.[Medline]

Piipari R, Savela K, Nurminen T, Hukkanen J, Raunio H, Hakkola J, Mantyla T, Beaune P, Edwards RJ, Boobis AR, et al. (2000) Expression of CYP1A1, CYP1B1 and CYP3A, and polycyclic aromatic hydrocarbon-DNA adduct formation in bronchoalveolar macrophages of smokers and non-smokers. Int J Cancer 86: 610–616.[CrossRef][Medline]

Quattrochi LC and Guzelian PS (2001) Cyp3A regulation: from pharmacology to nuclear receptors. Drug Metab Dispos 29: 615–622.[Abstract/Free Full Text]

Raunio H, Hakkola J, Hukkanen J, Lassila A, Paivarinta K, Pelkonen O, Anttila S, Piipari R, Boobis A, and Edwards RJ (1999) Expression of xenobiotic-metabolizing CYPs in human pulmonary tissue. Exp Toxicol Pathol 51: 412–417.[Medline]

Remacle JE, Kraft H, Lerchner W, Wuytens G, Collart C, Verschueren K, Smith JC, and Huylebroeck D (1999) New mode of DNA binding of multi-zinc finger transcription factors: deltaEF1 family members bind with two hands to two target sites. EMBO J 18: 5073–5084.[CrossRef][Medline]

Ringrose L and Paro R (2004) Epigenetic regulation of cellular memory by the polycomb and trithorax group proteins. Annu Rev Genet 38: 413–443.[CrossRef][Medline]

Rodríguez-Antona C, Bort R, Jover R, Tindberg N, Ingelman-Sundberg M, Gomez-Lechon MJ, and Castell JV (2003) Transcriptional regulation of human CYP3A4 basal expression by CCAAT enhancer-binding protein alpha and hepatocyte nuclear factor-3 gamma. Mol Pharmacol 63: 1180–1189.[Abstract/Free Full Text]

Saito T, Takahashi Y, Hashimoto H, and Kamataki T (2001) Novel transcriptional regulation of the human CYP3A7 gene by Sp1 and Sp3 through nuclear factor B-like element. J Biol Chem 276: 38010–38022.[Abstract/Free Full Text]

Schuetz EG (2004) Lessons from the CYP3A4 promoter. Mol Pharmacol 65: 279–281.[Free Full Text]

Sekido R, Murai K, Kamachi Y, and Kondoh H (1997) Two mechanisms in the action of repressor deltaEF1: binding site competition with an activator and active repression. Genes Cells 2: 771–783.[Abstract]

Sheweita SA (2000) Drug-metabolizing enzymes: mechanisms and functions. Curr Drug Metab 1: 107–132.[Medline]

Stevens JC, Hines RN, Gu C, Koukouritaki SB, Manro JR, Tandler PJ, and Zaya MJ (2003) Developmental expression of the major human hepatic CYP3A enzymes. J Pharmacol Exp Ther 307: 573–582.[Abstract/Free Full Text]

Watkins PB (1994) Noninvasive tests of CYP3A enzymes. Pharmacogenetics 4: 171–184.[Medline]

Westlind-Johnsson A, Malmebo S, Johansson A, Otter C, Andersson TB, Johansson I, Edwards RJ, Boobis AR, and Ingelman-Sundberg M (2003) Comparative analysis of cyp3a expression in human liver suggests only a minor role for cyp3a5 in drug metabolism. Drug Metab Dispos 31: 755–761.[Abstract/Free Full Text]

Willey JC, Coy E, Brolly C, Utell MJ, Frampton MW, Hammersley J, Thilly WG, Olson D, and Cairns K (1996) Xenobiotic metabolism enzyme gene expression in human bronchial epithelial and alveolar macrophage cells. Am J Respir Cell Mol Biol 14: 262–271.[Abstract]

Xie HG, Wood AJ, Kim RB, Stein CM, and Wilkinson GR (2004) Genetic variability in CYP3A5 and its possible consequences. Pharmacogenomics 5: 243–272.[CrossRef][Medline]

Yeh KT, Chen JC, Chen CM, Wang YF, Lee TP, and Chang JG (2003) CYP3A5*1 is an inhibitory factor for lung cancer in Taiwanese. Kaohsiung J Med Sci 19: 201–207.[Medline]

Zhang W, Purchio AF, Chen K, Wu J, Lu L, Coffee R, Contag PR, and West DB (2003) A transgenic mouse model with a luciferase reporter for studying in vivo transcriptional regulation of the human CYP3A4 gene. Drug Metab Dispos 31: 1054–1064.[Abstract/Free Full Text]

作者单位：Department of Pharmacology & Toxicology, University of Utah, Salt Lake City, Utah (J.S.B., J.W., N.S.C., G.S.Y.); Department of Pharmacology and Toxicology, University of Oulu, Oulu, Finland (J.H., P.U.); and University of Kuopio, Kuopio, Finland (H.R.)

【关键词】  N-Ethylmaleimide-Sensitive

    Recycling of G protein-coupled receptors determines the functional resensitization of receptors and is implicated in switching β2 adrenoceptor (β2AR) G protein specificity in cardiomyocytes. The human β2AR carboxyl end binds to the N-ethylmaleimide-sensitive factor (NSF), an ATPase integral to membrane trafficking machinery. It is interesting that the human β2AR (hβ2AR) carboxyl end pulled down NSF from mouse heart lysates, whereas the murine one did not. Despite this difference, both β2ARs exhibited substantial agonist-induced internalization, recycling, and Gi coupling in cardiomyocytes. The hβ2AR, however, displayed faster rates of agonist-induced internalization and recycling compared with the murine β2AR (mβ2AR) and a more profound Gi component in its contraction response. Replacing the mβ2AR proline (-1) with a leucine generated a gain-of-function mutation, mβ2AR-P417L, with a rescued ability to bind NSF, faster internalization and recycling than the mβ2AR, and a significant enhancement in Gi signaling, which mimics the hβ2AR. Selective disruption of the mβ2AR-P417L binding to NSF inhibited the receptor coupling to Gi. Mean-while, inhibiting NSF with N-ethylmaleimide blocked the mβ2AR recycling after agonist-induced endocytosis. Expressing the NSF-E329Q mutant lacking ATPase activity inhibited the mβ2AR coupling to Gi in cardiomyocytes. Our results revealed a dual regulation on hβ2AR trafficking and signaling by NSF through direct binding to cargo receptor and its ATPase activity and uncovered an unprecedented role for the receptor binding to NSF in regulating G protein specificity that has diverged between mouse and human β2ARs.

    β-Adrenoceptors play a pivotal role in regulating cardiomyocyte contraction through distinct signaling pathways. The β1AR couples to Gs protein(s), which increases cAMP/protein kinase A activity and the contraction rate, whereas the activated β2AR sequentially couples to both Gs and Gi in neonatal cardiomyocytes, creating a biphasic change in contraction. β2AR Gi coupling seems to be dependent on receptor trafficking, which includes both endocytosis and recycling. Inhibiting either process blocks receptor coupling to Gi in cardiomyocytes (Xiang et al., 2002; Xiang and Kobilka, 2003).

    Many G protein-coupled receptors (GPCRs) undergo endocytosis in response to activation, yet their subsequent sorting in endosomes is variable, creating variable regulation of their activity during prolonged or repeated stimulation. Some receptors are targeted to lysosomes to down-regulate cellular responses mediated by the receptor, whereas many GPCRs possess the ability to efficiently return to the cell surface. This recycling of receptors underlies the resensitization of corresponding cellular responses (von Zastrow, 2003). Many GPCRs depend on sequences residing in their intracellular domains for recycling. A well-defined class of recycling sequences are PSD-95/Discs-large/ZO-1 (PDZ) domain binding motifs (also called PDZ ligands) that are usually located at the carboxyl-terminal end of different GPCR tails (Bockaert et al., 2004; Gage et al., 2005). The β2AR has a type I PDZ ligand at its carboxyl-terminal end that is necessary for recycling and sufficient to reroute the -opioid receptor from a degradative to a recycling pathway (Cao et al., 1999; Gage et al., 2001). In cultured neonatal mouse cardiomyocytes, this sequence is also required for the temporal switch from Gs to Gi-mediated signal transduction observed in the contraction-rate response to the agonist isoproterenol (Xiang and Kobilka, 2003). Several lines of evidence now indicate that membrane trafficking of this receptor dictates not only cellular resensitization but also signal transduction specificity. Despite progress in understanding the β2AR recycling process, numerous questions concerning the core mechanism and physiological variations remain.

    Although the recycling sequence at the β2AR C terminus has been shown to bind PDZ domains in NHERF family proteins (NHERF-1/EBP50 and NHERF-2/E3KARP) (Hall et al., 1998; Cao et al., 1999), it also binds at least one protein with no identifiable PDZ domain: the N-ethylmaleimide sensitive factor (NSF) (Cong et al., 2001). NSF has been identified as an ATPase that binds SNAP receptor (SNARE) complexes in an ATP-dependent fashion to separate them during ATP hydrolysis; this and a wealth of other evidence has demonstrated its general role in vesicle fusion between various membrane compartments (Morgan and Burgoyne, 2004; Whiteheart and Matveeva, 2004). Moreover, NSF has been shown to bind to β-arrestin, an adaptor protein involved in GPCR desensitization and endocytosis upon agonist stimulation. β-Arrestin preferentially interacts with the ATP-bound form of NSF, and this NSF binding facilitates clathrin coat-mediated GPCR internalization (McDonald et al., 1999). In heterologous HEK293 cells, selective ablation of NSF binding to the β2AR was inferred to inhibit recycling of receptors, whereas imparting NSF binding on the -opioid receptor slightly enhanced its ability to recycle (Cong et al., 2001; Gage et al., 2005). Although there is also evidence to show that PDZ interactions promote receptor recycling (Cao et al., 1997) and are functionally important for Gi coupling in cardiomyocytes (Xiang and Kobilka, 2003), it is not clear how NSF may affect β2AR trafficking and signaling in these cells.

    Here, we used neonatal mouse cardiomyocytes as a model system to address these questions. It is interesting that the NSF binding sites on the β2AR were not conserved among mammalian species, providing a naturally occurring divergence in NSF binding to exploit. The -1 position of the β2AR carboxyl terminus is proline in mβ2AR and leucine in hβ2AR. Because of this single amino acid difference, mβ2AR binding to NSF was not detectable. Nevertheless, despite the lack of detectable binding of the mβ2AR carboxyl terminus to NSF in biochemical assays, we found that inhibition of NSF activity with N-ethylmaleimide (NEM) inhibited murine β2AR (mβ2AR) recycling despite this poor affinity. In addition, both human and murine β2ARs sufficiently recycled after endocytosis and coupled to Gi pathways in cardiomyocytes. The different affinities for NSF seemed to have a minimum role on receptor trafficking and signaling. In contrast, inactivation of NSF ATPase activity with a point mutation was sufficient to block both human and murine β2AR recycling and coupling to Gi in cardiomyocytes, indicating that NSF is required for proper trafficking and signaling of β2ARs in cardiomyocytes independent of a high-affinity interaction with the receptor. This study strengthens the relationship between β2AR recycling and signaling specificity and demonstrates an unprecedented role for NSF in regulating physiologically relevant signal transduction.

    cDNA Constructs and Mutagenesis. Constructs containing the cloned human and murine β2AR in pcDNA3 (Invitrogen, Carlsbad, CA) with a FLAG epitope attached at the N terminus were used for these studies and have been described before (Cao et al., 1999; Swaminath et al., 2004). Constructs encoding for GST-β2AR and GST-β2AR-alanine proteins (encompassing amino acids 328 to 413 of the human β2AR, and the latter with an additional alanine added to the C terminus) have also been reported (Cao et al., 1999). A comparable murine GST-β2AR construct was created by insertion of a polymerase chain reaction product of the region encoding amino acids 328 to 418 of the FLAG-mβ2AR construct using primers containing EcoRI and HindIII appendages and performing the appropriate digestion and ligation into pGEX-KG (Pfizer, New York, NY). The human NSF coding sequence was similarly ligated into pEGFP-N1 (Clontech, Mountain View, CA) after SacI digestion of the vector and a polymerase chain reaction product containing a 3'-SacI appendage and including the 5'-SacI restriction site from the source vector, NSF in pBluescriptR (American Type Culture Collection, Manassas, VA). The P417L mutation was introduced into the FLAG-mβ2AR and GST-mβ2AR constructs via the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA), as was the E329Q NSF mutation into the pEGFP-N1 construct. Plasmid amplification was done in DH5 Escherichia coli, and all sequences were verified by dideoxynucleotide sequencing (University of California San Francisco Biomolecular Resource Center, San Francisco, CA).

    Cell Culture and Transfection. Spontaneously beating neonatal cardiomyocytes were prepared from hearts of 1-day-old β1/β2AR-KO mouse pups as before (Devic et al., 2001). The myocyte-enriched cells remaining in suspension after preplating were plated in 35-mm dishes for contraction-rate studies and in 12-well plates for immunological assays (with coverslips for immunofluorescent microscopy). Recombinant adenovirus encoding FLAG-mβ2AR has been described previously (Xiang et al., 2002), and the FLAG-mβ2AR/P417L, FLAG-hβ2AR, GFP-NSF, and GFP-NSF-E329Q adenoviral vectors were generated with the same pAdEasy system (Qbiogene Inc., Irvine, CA). Neonatal myocytes were infected with viruses at a multiplicity of infection of 100 after being cultured for 24 h. The receptor expression levels were determined by ligand binding assays as described previously (Xiang et al., 2002). They were expressed at equivalent levels in cardiac myocytes (FLAG-mβ2AR, 147.3 ± 22 fmol/mg; FLAG-mβ2AR/P417L, 171.6 ± 9.1 fmol/mg; and FLAG-hβ2AR, 160.3 ± 21.8 fmol/mg membrane).

    GST Pulldown Assays. The various GST-β2AR fusion proteins were produced in BL21 E. coli and bound to glutathione-Sepharose agarose beads (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK). Beads containing 10 µg of the full-length fusion protein (assessed by densitometry of Coomassie-stained protein resolved by SDS-polyacrylamide gel electrophoresis) were incubated for 4 h at 4°C in 0.5 ml of clarified extracts from frozen mouse hearts with atria removed (Pel-Freez Biologicals, Rogers, AR), prepared to 10 mg/ml. Beads were washed four times in 1 ml of extract buffer [0.1% (v/v) Triton X-100, 150 mM NaCl, 25 mM KCl, and 10 mM Tris, pH 7.4, complete Roche protease inhibitor cocktail], and protein was eluted in lithium dodecyl sulfate sample buffer (Invitrogen) with dithiothreitol added to 20 mM. Samples were divided in two for SDS-polyacrylamide gel electrophoresis, transfer to nitrocellulose, and Western blotting using rabbit anti-EBP50 antibodies (courtesy of Dr. Anthony Bretscher, Cornell University, Ithaca, NY) or the mouse 2E5 anti-NSF antibody (courtesy of Dr. Sidney W. Whiteheart, University of Kentucky, Lexington, KY).

    Immunofluorescence Microscopy. Myocyte images were obtained using a similar setup on a Zeiss Axioplan 2 microscope (Carl Zeiss Inc., Thornwood, NY). Fluorescent measurements of the myocyte receptor trafficking were made by a ratiometric normalization of fluorescent intensities measured using Metamorph software (Molecular Devices, Sunnyvale, CA). Epitope-tagged receptors were detected using M1 anti-FLAG antibody (Sigma, St. Louis, MO). Selective detection of surface relative to total pools of receptor and its use to estimate receptor recycling have been described previously (Tanowitz and von Zastrow, 2003). The recycling estimates were conducted without the EDTA strip. The primary antibody used in these experiments was M1 conjugated to Alexa Fluor 488 (Invitrogen) using standard procedures as described previously (Tanowitz and von Zastrow, 2003). Secondary staining was performed using a commercial goat antimouse IgG Alexa Fluor 594 conjugate (Invitrogen). Experiments were performed at least in triplicate, and representative results are shown.

    Fig. 1. The binding of mβ2AR and hβ2AR to NSF and NHERF-1/EBP50. The binding of mβ2AR and hβ2AR to NSF and NHERF-1/EBP50 from mouse heart extracts. GST pull-downs were performed as described under Materials and Methods. Western detection of NHERF-1/EBP50 pulled down by the indicated GST-β2AR fusion protein from mouse heart extract, and the corresponding detection of NSF is shown in A, whereas the detection of NHERF-1/EBP50 from this extract is shown in B. Ponceau-stained GST fusion proteins are shown under each blot. Images are representative of three or more experiments.

    Immunofluorescence Spectroscopy. Surface receptor levels were determined as before (Swaminath et al., 2004) in the indicated cell type expressing the indicated FLAG-β2AR. Media were refreshed 1 h before 10 µM isoproterenol (Sigma) stimulation for 10 or 30 min. Periods of agonist washout after 30-min isoproterenol stimulations were also performed for an additional 30 or 60 min as indicated.

    Myocyte Contraction Rate Assay. Measurement of spontaneous contraction rates from myocytes expressing either the endogenous or the indicated FLAG-β2AR were carried out with and without the use of PTX as described previously (Devic et al., 2001). In some assays, NEM was applied 30 min before the addition of isoproterenol. Tat peptide, Tat-β2-DSAL consisting of Tat linked to GRQGFSSDSAL of β2AR, and Tat-β2-ASLL consisting of Tat linked to GRQGFSSASLL of β2AR through a cysteine bridge were synthesized in the Stanford Core facility and EZ-Biolab (Indianapolis, IN). Neonatal myocytes were preincubated at 37°C with 10 µM peptide for 25 min before isoproterenol (10 µM; Sigma) exposure.

    Statistical Analysis. Curve-fitting and statistical analyses were performed using Prism (GraphPad Software, Inc., San Diego, CA).

    NSF Had Higher Binding Affinity to Human β2AR than Murine β2AR. To understand the molecular mechanism of the NSF effect on β2AR signaling in cardiomyocytes, the interaction between β2AR and NSF from heart lysate was examined. NSF, a hexameric ATPase involved in membrane fusion, can bind to the carboxyl terminus of the hβ2AR. The protein-binding region on this receptor involves a four-residue stretch at the distal C terminus of the receptor (Cong et al., 2001). In addition, NHERF-1/EBP50, a cytoskeleton-associated protein, can also bind to the same stretch of residues on the carboxyl terminus of the hβ2AR. It is interesting that several rodent β2ARs, including the mβ2AR, are identical with the hβ2AR in the carboxyl-binding region except at one residue (leucine -1 of the hβ2AR, DSLL) that is required for binding to NSF but not to NHERF-1/EBP50 (Cong et al., 2001). Rather, the mβ2AR has a proline at the -1 position (DSPL). To determine whether both the human and murine β2ARs have a similar capacity to bind NSF, GST-fusion proteins, including various β2AR carboxyl-terminal tail sequences, were prepared. Protein binding was evaluated with a pull-down assay using the GST-fusion proteins coupled to glutathione-agarose beads.

    GST fusion proteins were incubated with tissue lysate prepared from mouse hearts. NSF only bound to the cytoplasmic tail of the hβ2AR but not the mβ2AR (Fig. 1A). In contrast, NHERF-1/EBP50 bound to the cytoplasmic tail of both the hβ2AR and mβ2AR under these conditions (Fig. 1B). As a negative control for nonspecific binding, an addition of a single alanine residue to the hβ2AR carboxyl terminus (GSThβ2AR-Ala) was tested as well. As reported previously, this mutant failed to exhibit the PDZ domain-mediated and NSF protein binding (Fig. 1, A and B; Cao et al., 1999). In addition, we attempted to "rescue" an NSF interaction with the mβ2AR tail by substitution of the mβ2AR proline 417 with a leucine residue (mβ2AR-P417L). The mβ2AR-P417L pulled down similar amounts of NSF from lysates compared with the hβ2AR, indicating that the mβ2AR-P417L cytoplasmic tail fully rescued binding to NSF (Fig. 1A). Likewise, this mutant mβ2AR-P417L pulled down NHERF-1/EBP50 from mouse heart lysates (Fig. 1B).

    The Binding of NSF and the NSF ATPase Activity Had Distinct Effects on β2AR Trafficking in Cardiomyocytes. To examine whether NSF plays any role in β2AR trafficking in cardiomyocytes, we analyzed the localization of flag-tagged β2ARs in cardiac myocytes. The mβ2AR, mβ2AR-P417L, and hβ2AR were transiently expressed in cardiac myocytes using recombinant adenovirus. Immunofluorescence studies showed that all three receptors had a cell-surface staining during a nonstimulated state (Fig. 2A). Upon isoproterenol stimulation, all three receptors had reduced cell-surface staining together with increased punctate intracellular staining, suggesting a significant internalization of the receptors in cardiac myocytes. These observations were confirmed quantitatively using an ELISA-based method for assaying surface receptor levels (Swaminath et al., 2004) in a large number of cells and a ratiometric method for analysis of fluorescence micrographs (Tanowitz and von Zastrow, 2003) (Fig. 2B; data not shown). It is interesting that when we measured the short-term decrease in cell-surface receptors after agonist stimulation in cardiac myocytes, we found that the mβ2AR-P417L had a faster rate (t = 2.63 ± 0.05 min) of cell surface-receptor decrease than the mβ2AR (t = 10.93 ± 0.01 min; Fig. 3). Because the receptor level change in the short time points after agonist stimulation is primarily determined by agonist-induced endocytosis, these data suggested a faster rate of endocytosis for the mβ2AR-P417L than for the mβ2AR in cardiac myocytes. The hβ2AR had a similar rate (t = 3.4 ± 0.03 min) of cell surface-receptor decrease compared with the mβ2AR-P417L (Fig. 3). However, after 30 min of agonist stimulation, we observed a similar amount of surface receptor decreases with the mβ2AR (30.54 ± 2.90%), the mβ2AR-P417L (28.43 ± 1.69%), and the hβ2AR (24.94 ± 2.74%). The observed decrease of receptor density at 30 min of stimulation should have been a composite of receptor endocytosis and recycling. The equivalent decreases of receptors at cell surface are usually due to much slower recycling process than endocytosis in cells.

    Fig. 2. NSF binding enhances recycling of FLAG-β2ARs in neonatal cardiac myocytes from β1/β2AR-KO mice. A, human and murine β2ARs internalize and recycle in cardiac myocytes. Cardiac myocytes expressing a FLAG-tagged hβ2AR, mβ2AR, or mβ2AR-P417L were stained with M1 primary antibody conjugated to the Alexa-488 fluorophore to observe a starting "total" receptor population. After no treatment (0), 30-min 10 µM isoproterenol treatment (30), or 30-min isoproterenol treatment followed by a surface antibody strip and 60 min of agonist removal (30 + 60), cells were stained under nonpermeable conditions with a goat anti-mouse-IgG secondary antibody conjugated to the Alexa-594 fluorophore to observe the relative complement of "surface" receptor. Images are representative of three experiments. B, NSF binding β2ARs recycle faster. Surface levels of the three β2ARs were quantified by fluorescence spectroscopy measurements of M1-Alexa 488 associated with the cell surface receptors after the indicated periods of drug administration and removal (1, control; 2, 30 min of isoproterenol stimulation; 3, 30 min of isoproterenol followed by 30 min of drug removal; and 4, 30 min of isoproterenol followed by 60 min of drug removal). Surface levels are normalized as a percentage of untreated cell surface fluorescence, and error bars reflect standard deviations over three experiments. *, p < 0.05, significantly different between mβ2AR and hβ2AR or mβ2AR-P417L by t test.

    Fig. 3. NSF binding enhances the internalization kinetics of FLAG-β2ARs expressed in neonatal cardiac myocytes from β1/β2AR KO mice. Surface levels of the three β2ARs were quantified by fluorescent measurement of M1-Alexa 488 associated with the cell surface receptors after the indicated periods of 10 µM isoproterenol administration. Data were normalized as a percentage decrease of untreated cell surface fluorescence, and error bars reflect standard deviations over three experiments. The data represent the mean ± S.E. of experiments from at least three different myocyte preparations.

    When isoproterenol was removed, both the hβ2AR and mβ2AR-P417L recovered cell-surface staining almost completely after a 60-min incubation (Fig. 2). In contrast, the mβ2AR did not show a fully recovered cell-surface staining pattern, and some residual intracellular staining was observed in these cells (Fig. 2), even though the majority of the internalized receptors seemed to return to the surface within 60 min. When we examined the cell surface-receptor density at different time points with the fluorescent ELISA assay, the mβ2AR exhibited a lower recovery of cell surface-receptors after recycling for 60 min than the mβ2AR-P417L mutant and the hβ2AR (Fig. 2B; *, p < 0.05). A significant difference in surface recovery was also observed using ratiometric image measurements 60 min after agonist washout (data not shown). These data indicate that the mβ2AR, although capable of undergoing agonist-induced internalization and recycling in cardiomyocytes, differs in rates of recycling compared with the hβ2AR and mβ2AR-P417L.

    It has been well-established that NSF ATPase activity plays an important role in membrane cargo trafficking. We then tested whether NSF activity was necessary for the endocytic recycling of the receptor by using NEM to inhibit NSF activity in myocytes. In the presence of NEM, endocytosis of the receptor was preserved; however, a return of the receptor to the cell surface after removal of agonist for 60 min was not (Fig. 4A). This observation was confirmed with measurements of surface receptor levels by a fluorescent ELISA assay. The cell surface receptor levels dropped after agonist addition and only recovered with agonist withdrawal in the absence of NEM (Fig. 4B). These data suggested that NEM treatment can block the receptor from recycling after endocytosis.

    Fig. 4. Inhibiting NSF with NEM blocks the FLAG-β2AR recycling after agonist-induced endocytosis in cardiomyocytes. A, murine β2ARs internalize and recycle in cardiac myocytes. Cardiac myocytes expressing a FLAG-tagged mβ2AR were treated as described under Materials and Methods and Fig. 3. Images are representative of three experiments. Inhibiting NSF with NEM blocks the FLAG-β2AR recycling after agonist-induced endocytosis in cardiomyocytes. B, surface levels of the mβ2ARs were quantified by fluorescent measurement of M1-Alexa 488 associated with the cell surface receptors after the indicated periods of drug administration and removal (1, control; 2, 30 min of isoproterenol stimulation; 3, 30 min of isoproterenol followed by 30 min of drug removal; and 4, 30 min of isoproterenol followed by 60 min of drug removal). Surface levels were normalized as a percentage of untreated cell surface fluorescence, and error bars reflect standard deviations over three experiments. *, p < 0.05, significantly different between cells with and without NEM treatment by t test.

    Dominant-Negative NSF Lacking ATPase Activity Inhibited Endogenous mβ2AR Coupling to Gi Pathway in Cardiomyocytes. Our finding that NEM inhibits β2AR recycling in cardiac myocytes suggested that NSF function is required for this process. To further probe whether NSF enzymatic activity can affect the receptor signaling independent from the direct NSF-receptor interaction, we examined the signaling mediated by the endogenous mβ2AR when overexpressing an inactivated NSF, the E329Q mutant (Whiteheart et al., 1994). This mutation abolishes ATPase activity and has been shown to block AMPA receptor trafficking (Whiteheart et al., 1994; Whiteheart and Matveeva, 2004). When the endogenous mβ2AR in the β1AR-KO myocyte was stimulated by isoproterenol, the activated receptor induced a biphasic contraction-rate response with an initial increase mediated by Gs coupling followed by a sustained Gi-dependent decrease to reduce the contraction rate below basal level (Fig. 5A; Xiang et al., 2002). When wild-type NSF was expressed in β1AR-KO cardiac myocytes, we did not observe any significant change in the endogenous mβ2AR-mediated contraction-rate response (Fig. 5A). In contrast, when the NSF-E329Q mutant was overexpressed in cardiomyocytes, the contraction rate mediated by the mβ2AR was significantly higher than the control and did not display a decrease lower than the basal level (Fig. 5B). This response profile was similar to that observed with an inhibition of Gi by PTX (Fig. 5C). Indeed, additional treatment of PTX did not generate any further increases in contraction rates (Fig. 5D). Therefore, the NSF-E329Q behaved as a dominant-negative to block the receptor coupling to Gi in cardiomyocytes. In addition, when myocytes are pretreated with NEM to inhibit the NSF ATPase activity, we also observed effects similar to those by NSF-E329Q mutant on mβ2AR signaling mediated contraction-rate response (data not shown).

    Fig. 5. Dominant-negative NSF-E329Q mutant inhibits the mβ2AR coupling to Gi protein. Spontaneously beating cardiac myocytes from β1AR KO mice were transfected with a wild-type NSF (A and C) or NSF-E329Q (B and D) mutant adenovirus as indicated. The cells were administered 10 µM isoproterenol with inhibition of Gi by PTX. Overexpressing the NSF E329Q mutant enhanced the contraction-rate increase induced by isoproterenol stimulation. Additional PTX treatment did not further enhance the contraction-rate increase induced by the mβ2AR. The data represent the mean ± S.E. of experiments from at least three different myocyte preparations. *, p < 0.05, time course significantly different by two-way ANOVA.

    The Divergent C Termini of the Human and Murine β2AR Had Different Effects on Contraction Rate Responses in Neonatal Cardiomyocytes. Our previous studies have shown that the localization and trafficking of the mβ2AR is important for the receptor's G protein signaling specificity and subsequent regulation of the myocyte contraction rate. In the course of this study, we found that the divergent PDZ ligand of the human and murine β2AR affected the receptor trafficking rates after agonist stimulation in cardiac myocytes. Thus, we wanted to examine whether differences in NSF binding and/or altered trafficking rates could modulate the receptor signaling in cardiac myocytes. When the mβ2AR was expressed in β1/β2AR-KO myocytes and stimulated by isoproterenol, the activated receptor induced a biphasic contraction-rate response with an initial increase followed by a sustained decrease to reduce the contraction rate lower than basal level (Fig. 6A; Xiang et al., 2002). This contraction-rate change is equivalent to that induced by the endogenous mβ2AR in β1AR-KO myocytes (Fig. 6A). The mβ2AR-P417L induced a similar contraction-rate response profile and initial increase compared with the mβ2AR in β1/β2AR-KO myocytes (Figs. 6C and 7D). However, the contraction rate decreased faster, and the contraction rate was lower than that induced by the mβ2AR during late stimulation in cardiac myocytes (Figs. 6C and 7E). In addition, when stimulating the hβ2AR expressed in β1/β2AR-KO myocytes with isoproterenol, the activated receptor also induced a biphasic, contraction-rate change with an initial increase followed by a sustained decrease (Fig. 6B). Although it is interesting that the initial contraction-rate increase was smaller than that induced by the activated mβ2AR and mβ2AR-P417L (Figs. 6B and 7D), it is more surprising that the late decrease in contraction rate induced by the exogenous hβ2AR was greater than that induced by the mβ2AR and mβ2AR-P417L (Figs. 6B and 7E).

    Fig. 6. Differences in β2AR contraction-rate responses to isoproterenol in neonatal cardiac myocytes from β1/β2AR KO mice. The hβ2AR and mβ2AR-P417L exhibit different contraction rate profiles than the mβ2AR at comparable expression levels. Spontaneously beating, cardiac myocytes from β1/β2AR KO mice were infected with a FLAG-tagged mβ2AR (A), hβ2AR (B), or mβ2AR-P417L (C) recombinant adenovirus as indicated and infused with 10 µM isoproterenol. Contraction rates were measured and normalized as the change over baseline. The data represent the mean ± S.E. of experiments from at least three different myocyte preparations. *, p < 0.05, time course significantly different by two-way ANOVA.

    Fig. 7. NSF binding enhances the Gi signaling components of β2ARs. Spontaneously beating cardiac myocytes from β1/β2AR KO mice were transfected with a FLAG-tagged mβ2AR (A), mβ2AR-P417L (B), or hβ2AR (C) adenovirus as indicated. The cells were administered 10 µM isoproterenol with inhibition of Gi with PTX. PTX treatment did not affect initial response usually mediated by receptor/Gs coupling (D) but significantly enhanced the contraction rate during the late stimulation induced by the Gi coupling to the activated hβ2AR, mβ2AR, or mβ2AR-P417L (E). The data represent the mean ± S.E. of experiments from at least three different myocyte preparations. *, p < 0.05, time course significantly different by two-way ANOVA. **, p < 0.05, unpaired t test significantly different on initial maximum contraction rate increases or late contraction-rate decreases mediated by different β2ARs. ***, p < 0.05, unpaired t test significantly different on late contraction rate decreases after PTX treatment.

    The profound contraction-rate decrease induced by the mβ2AR-P417L and the hβ2AR suggests that these receptors may have enhanced coupling to Gi and/or reduced coupling to Gs compared with the mβ2AR. We therefore examined the Gi signaling induced by the activated receptors in cardiac myocytes. PTX was used to block Gi signaling in cardiac myocytes expressing the different β2ARs before isoproterenol stimulation. Upon inhibiting Gi with PTX, the isoproterenol-stimulated mβ2AR induced a slightly greater but not significant contraction-rate increase in myocytes compared with the control and prevented the late Gi-dependent contraction rate decrease (Fig. 7A; Xiang et al., 2002). PTX treatment also inhibited the contraction rate decrease mediated by the mβ2AR-P417L or the hβ2AR during the late phase of stimulation (Fig. 7, B and C). These data suggest that compared with the activated mβ2AR, the mβ2AR-P417L had an enhanced Gi coupling upon isoproterenol stimulation, and the activated hβ2AR coupled to Gi more efficiently in neonatal cardiac myocytes (Fig. 7E). This indicates that the divergent receptor C termini can induce different changes in contraction-rate responses that correlate with subtle changes in receptor transportation rates.

    The Binding of NSF and PDZ Had Distinct Effects on β2AR Activation-Induced Contraction Rates in Cardiomyocyte. To further probe the effect of the β2AR binding to NSF and PDZ on receptor signaling in cardiomyocytes, we took advantage of the different binding affinities between receptor and proteins by using peptides to selectively disrupt the interactions. We expressed either mβ2AR or mβ2AR-P417L (the hβ2AR mimic) in β1/β2AR-KO cardiomyocyte for the contraction rate assay. Membrane-permeable peptides containing ASLL sequence and DSAL sequence were used to selectively disrupt NSF and NHERF/EBP50 binding, respectively. When mβ2AR-expressing myocytes were treated with NSF (ASLL) peptide, the activated receptor induced a slightly bigger but not significant initial increase than the cells without pretreatment (Fig. 8, A and C). The increase was sustained during stimulation and lacked a late decrease mediated by receptor/Gi coupling in control cells (Fig. 8, A and D). When mβ2AR-expressing myocytes were treated with PDZ (DSAL) peptide, the activated receptor induced a significantly greater initial increase than the control (Fig. 8, B and C), and the increase was sustained and lacked a late Gi-dependent decrease (Fig. 8, B and D).

    Fig. 8. Selective disruptions of NSF and NHERF-1/EBP50 binding have distinct effects on β2AR signaling. Spontaneously beating cardiac myocytes from β1/β2AR KO mice were transfected with a FLAG-tagged mβ2AR (A-D) or mβ2AR-P417L (E-H) adenovirus as indicated. The cells were administered 10 µM isoproterenol with pretreatment of membrane-permeable NSF peptide ASLL and PDZ peptide DSAL to disrupt the receptor binding to NSF and PDZ protein, respectively. NSF peptide ASLL significantly affected the receptor-mediated contraction response during the late stimulation, which are usually mediated by receptor/Gi coupling (D and H). In contrast, PDZ peptide affected both initial contraction rate increase mediated by receptor/Gs coupling (C and G) and the late contraction-rate response mediated by receptor/Gi coupling (D and H). The data represent the mean ± S.E. of experiments from at least three different myocyte preparations. *, p < 0.05, time course significantly different by two-way ANOVA. **, p < 0.05, unpaired t test significantly different on initial maximum contraction-rate increases or late contraction-rate decreases after treatment with peptides.

    In contrast, pretreatment with NSF (ASLL) peptide did not affect the activated mβ2AR-P417L-induced initial increase (Fig. 8, E and G). However, the increase was sustained and did not display a late Gi-induced decrease of contraction rate (Fig. 8, E and H). When myocytes expressing mβ2AR-P417L were treated with PDZ (DSAL) peptide, the activated receptor induced a significantly greater initial increase in contraction rate than the control (Fig. 8, F and G), and the increase was sustained and did not show a late Gi-induced decrease (Fig. 8, F and H). Together, these data showed that although disrupting the binding to PDZ protein (such as NHERF/EBP50) affects the receptor coupling to both Gs and Gi, disrupting the binding to NSF selectively affects the receptor coupling to Gi in cardiomyocytes.

    In the present study, several approaches were used to test whether NSF regulates β2AR trafficking and physiological signaling. This idea was extended from the studies of β2AR-selective interactions with NSF and PDZ proteins. A distal portion of the cytoplasmic C terminus of the hβ2AR selectively binds to several PDZ domain-containing proteins, such as the cytoskeleton-associated protein NHERF/EBP50, which is implicated in receptor recycling (Cao et al., 1999). However, a subsequent study confirmed the importance of the PDZ ligand for receptor recycling to the cell surface but identified a distinct non-PDZ interaction of this sequence with NSF that was required for proper endocytic recycling (Cong et al., 2001). Although the reported difference could result from the difference between the derived HEK293 cell lines, we have tried to address the functional roles of these receptor-protein interactions in primary cultured cardiomyocytes—a native environment that may have a more precise regulation of receptor function. We have shown previously that the carboxyl-terminal sequence of mβ2AR was also required for efficient plasma membrane recycling and for receptor coupling to Gi in cardiomyocytes (Xiang and Kobilka, 2003). In this study, we showed that the binding to NSF enhanced both internalization and recycling rates of β2AR and increased the receptor coupling to Gi signaling in cardiomyocytes (Figs. 2, 3, and 8). We further distinguished the effects of NSF and PDZ binding on β2AR signaling in myocytes. Although the binding to NSF increases receptor/Gi coupling, the binding to PDZ proteins affects receptor coupling to both Gs and Gi proteins (Fig. 8).

    It is interesting that, at the receptor's distal carboxyl terminus, the mβ2AR (DSPL) differs from the hβ2AR (DSLL) at the -1 position, at which the hβ2AR has a leucine critical for binding to NSF. Because the mβ2AR has a proline residue at the same relative position of the receptor cytoplasmic tail, we predicted that the receptor could not bind to NSF. Our experiments confirmed a very low affinity binding of NSF to the mβ2AR cytoplasmic tail (Fig. 1). We used a gain-of-function approach by replacing the proline with a leucine to generate a mutant mβ2AR-P417L. This mutant has a distal terminus identical with that of the hβ2AR (hβ2AR mimic) and displayed recovered binding to NSF (Fig. 1). The direct NSF-binding seemed to increase the rates of both agonist-induced endocytosis and recycling of the mβ2AR-P417L in cardiomyocytes (Fig. 2 and 3). We cannot exclude the possible contribution by the small increase in binding affinities of the mutant mβ2AR-P417L for PDZ proteins or new binding partners. However, our beating assay data supported that the increased trafficking rates are probably caused by the fact that the mutant mβ2AR-P417L gained binding to NSF (Fig. 8).

    Consistent with the trafficking data, the mβ2AR-P417L and the hβ2AR also displayed a more profound coupling to Gi than the mβ2AR in cardiomyocytes (Fig. 7). We have established previously that activated mβ2AR undergo sequential coupling to Gs and Gi to modulate cardiomyocyte contraction rate, and the recycling of mβ2AR is necessary for coupling to Gi (Xiang and Kobilka, 2003). By using membrane-permeable peptides to selectively inhibit the receptor binding to NSF or PDZ proteins, we will be able to distinguish the subtle effects of a specific binding on receptor signaling. Although disruption of PDZ binding affects receptor coupling to both Gs and Gi, disruption of NSF binding selectively inhibits receptor coupling to Gi (Fig. 8). It is interesting that despite that the mβ2AR does not bind to NSF well, the NSF peptide ASLL affected the receptor signaling (Fig. 8A). This may result from a low basal interaction between the mβ2AR and NSF. On the other hand, NSF peptide ASLL is capable of binding to PDZ proteins (Cong et al., 2001); thus, it may compete against DSPL on the mβ2AR, which is not a perfect PDZ ligand because of the structure of proline. In contrast, the binding between DSLL of the mβ2AR-P417L and PDZ proteins is less likely to be affected by the NSF peptide (Fig. 8E). Thus, the effect on the mβ2AR-P417L signaling by the NSF peptides suggested that the binding to NSF affects the receptor/Gi coupling, which is consistent with its role in the modulation of receptor recycling.

    NSF was identified as an ATPase, binding to SNARE complexes required for membrane fusion, thus playing critical roles in protein trafficking of many membrane receptors (Whiteheart and Matveeva, 2004). In agreement, we showed that NSF ATPase activity was essential for mβ2AR trafficking and signaling in cardiomyocytes (Fig. 4 and 5). It is interesting that NSF can bind to β-arrestin, an adaptor-like protein linking most GPCRs to clathrin-coated vesicles for endocytosis (McDonald et al., 1999). NSF binding to β-arrestin, like binding to classic SNARE substrates, is an ATP-dependent event (McDonald et al., 1999). Thus, NSF could play a role together with β-arrestin in recruiting the cargo receptors into clathrin-coated vesicles for budding. This process can be fine-tuned if NSF directly binds membrane cargo receptors, including hβ2AR (Heydorn et al., 2004). In addition, the binding of NSF to the hβ2AR is enhanced in the ATP-bound form (Gage et al., 2005), and the NSF ATPase activity dissociates the AMPA receptor from PDZ proteins allowing endocytosis (Osten et al., 1998; Hanley et al., 2002). Therefore, NSF can facilitate receptor recruitment into clathrin-coated vesicle by both direct binding to the cargo receptor and its ATPase activity. In the case of mβ2AR, the activated receptor recruits β-arrestin; this brings NSF to the receptor. NSF ATPase activity helps to dissociate the receptor from PDZ proteins to enter clathrin-coated vesicles, and later NSF regulates the vesicle fusion to endosome. In comparison, hβ2AR can directly bind to the NSF. When NSF is recruited to the receptor/arrestin complexes, it can compete against the receptor binding to PDZ proteins. This competition can lead to an increase in internalization rates (Fig. 3). During the receptor recycling, NSF binding can bridge the cargo receptor to SNARE complexes, which facilitate the docking of recycling vesicles to plasma membrane, hence enhancing the recycling rate of hβ2AR and mβ2AR-P417L but not mβ2AR. In this study, we only measured the cell surface receptor level during endocytosis and recycling. Any additional role of NSF in receptor trafficking among endosomal compartments remains to be addressed.

    It is noteworthy that the subtle effects of NSF-hβ2AR binding on trafficking and signaling is not conserved throughout mammals; such an effect can be overlooked easily in an experimental procedure. Although NSF is a common factor involved in membrane receptor trafficking, the context of the NSF-receptor complex can further complicate the type and degree of receptor regulation. These regulations will probably include the binding of the receptor to PDZ-domain containing proteins and cytoskeleton-associated proteins and additional binding of NSF to other trafficking proteins such as SNARE complexes and arrestin. Further studies using NSF mutants with selective ablation of binding to the β2AR or other proteins such as arrestin will help to dissect any roles of individual protein-protein interactions on the β2AR trafficking and signaling in cardiomyocytes or physiological settings.

    Indeed, an effect of NSF binding to the hβ2AR is likely to be complicated by competitive binding of PDZ proteins on the same sequence at the carboxyl-terminal end (Cong et al., 2001). A PDZ binding can have multiple effects on membrane receptor distribution and trafficking. One effect of PDZ binding is to stabilize and restrict the receptors at distinct subcellular domains. This is supported by the recent evidence that overexpressing NHERF-1/EBP50 reduced the agonist-induced internalization of two GPCRs, the parathyroid hormone receptor type-1 and thromboxane A(2)β receptor in HEK293 cells (Rochdi and Parent, 2003; Sneddon et al., 2003). By binding to cytoskeleton and/or scaffold proteins, the receptors can associate with signaling components and form complexes to either facilitate or restrict signal transduction. Consistent with the notion, our previous and current studies support that disrupting the PDZ binding to the β2ARs enhances the receptor coupling to Gs in cardiomyocytes (Xiang and Kobilka, 2003). In contrast, the PDZ protein GRIP/ABP binding has been shown to play a role in the stabilization of an intracellular pool of AMPA receptors that have been internalized with stimulation, thus inhibiting their recycling to the synaptic membrane (Braithwaite et al., 2002). Therefore, depending on the receptors and their binding partners, the PDZ domain-containing proteins can stabilize the receptor complexes at either the cell surface or intracellular compartments to fine-tune the receptor function in a given cell type. The third effect of PDZ binding seems to promote receptor trafficking to another subcellular location. Both PICK1 and NHERF-1/EBP50 have been shown to be critical for AMPA receptor and β2AR recycling back to the cell surface (Cao et al., 1999; Xiang and Kobilka, 2003; Lu and Ziff, 2005). In cardiomyocytes, selective disruption of PDZ binding with point mutations or with membrane-permeable peptide blocks receptor recycling and also inhibits receptor coupling to Gi ((Xiang and Kobilka, 2003) and Fig. 8). This PDZ-promoted trafficking may simply be a result of PDZ sequestration of receptors away from a competing trafficking fate, which could generalize PDZ interactions as hindrances to trafficking. The function of PDZ binding on receptor endocytosis and recycling could be further complicated by agonist-dependent phosphorylation of the receptor C-terminal end by G-protein receptor kinases and subsequent receptor dephosphorylation by pH-sensitive phosphatases (Sibley et al., 1986; Pitcher et al., 1995, 1998; Cao et al., 1999). The significance of this interplay in cardiomyocytes remains to be seen.

    When the hβ2AR was expressed in murine cardiomyocytes, the receptor displayed sequential coupling to Gs and Gi to regulate the myocyte contraction rate (Fig. 6). This result reinforced the notion from our previous studies that the recycling of the β2AR is part of a mechanism necessary for the receptor to switch from Gs to Gi (Xiang et al., 2002, 2005; Xiang and Kobilka, 2003). Both the human and murine β2ARs displayed a dual coupling to both Gs and Gi proteins in cardiac myocytes. Our studies revealed a species-dependent difference between human and murine β2ARs. The hβ2AR seemed to have a lower efficiency in coupling to the Gs pathway and a significantly higher efficiency in coupling to Gi than the mβ2AR when regulating the myocyte contraction rate (Figs. 6 and 7). Our results suggest that the profound Gi coupling is in part due to the increased binding to NSF. The mechanism of the low Gs coupling efficiency is not clear, although the higher receptor endocytosis rate could be an indication of enhanced desensitization. Another clue lies in the differences between receptor species. Despite the fact that the mβ2AR-P417L had recovered the ability to bind NSF, much of the signaling properties of this mutant mβ2AR still resembled those of the mβ2AR rather than the hβ2AR (Fig. 6). The differences of other structural domains on the hβ2AR and mβ2AR must thus account for the differences observed between the mβ2AR-P417L and the hβ2AR in cardiomyocytes. The notable regions include both the third loop and the proximal region of the carboxyl tail, which can directly influence G protein coupling. Another species-dependent difference is that a unique sugar-modification site located on the second extracellular domain of the hβ2AR, but not rodent β2ARs, promotes receptor degradation upon long-term agonist stimulation (Mialet-Perez et al., 2004). When overexpressed in mice, the hβ2AR seems to enhance the cardiac contraction in animal hearts without developing heart failure (Milano et al., 1995). The β2AR/Gs signaling is proapoptotic (Zhu et al., 2001), whereas the β2AR/Gi signaling plays an antiapoptotic role in both mouse hearts and cultured mouse cardiac myocytes (Zhu et al., 2001; Patterson et al., 2004). Thus, the more preferential coupling of the hβ2AR to Gi over Gs observed in our experiments could explain the lack of pathologic changes observed with overexpression of the hβ2AR in the hearts of mice. Further studies characterizing the differences between the hβ2AR and mβ2AR are needed to advance our understanding of adrenergic physiology in vivo.

    In conclusion, the present results indicate that NSF AT-Pase activity is necessary for agonist-dependent β2AR trafficking in cardiomyocytes, whereas NSF binding enhances the receptor transportation rates. Both the direct binding to NSF and its ATPase activity are important for the receptor coupling to Gi. Our data also showed different affinities of NSF binding to β2ARs from different species, and the direct binding to NSF contributes to the differences of receptor signaling in cardiomyocytes. Our data further revealed distinct effects of NSF and PDZ binding on β2AR signaling. In contrast to the selective effect on Gi coupling by the receptor binding to NSF, the receptor binding to PDZ proteins affects the receptor coupling to both Gs and Gi proteins. The present results add to the growing appreciation of diversified cellular factors as part of comprehensive mechanisms to fine-tune GPCR signaling and membrane trafficking in native mammalian cardiomyocytes.

    Acknowledgements

    We acknowledge Anthony Bretscher (Cornell) for anti-EBP50 and Sidney Whiteheart (University of Kentucky) for valuable discussion and for anti-NSF.

    ABBREVIATIONS: β2AR, β2-adrenergic receptor; β1AR, β1-adrenergic receptor; GPCR, G protein-coupled receptor; KO, knockout; NSF, N-ethylmaleimide-sensitive factor; PDZ, PSD-95/Discs-large/ZO-1; NHERF/EBP50, Na+/H+ exchanger regulatory factor/ezrin/radixin/moesin-binding phosphoprotein of 50 kDa; GST, glutathione; PTX, pertussis toxin; NEM, N-ethylmaleimide; ELISA, enzyme-linked immunosorbent assay; HEK, human embryonic kidney; ANOVA, analysis of variance; AMPA, -amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid; SNARE, soluble N-ethylmaleimide-sensitive factor attachment protein receptor.

【参考文献】
  Bockaert J, Dumuis A, Fagni L, and Marin P (2004) GPCR-GIP networks: a first step in the discovery of new therapeutic drugs? Curr Opin Drug Discov Dev 7: 649-657.[Medline]

Braithwaite SP, Xia H, and Malenka RC (2002) Differential roles for NSF and GRIP/ABP in AMPA receptor cycling. Proc Natl Acad Sci U S A 99: 7096-7101.[Abstract/Free Full Text]

Cao TT, Deacon HW, Reczek D, Bretscher A, and von Zastrow M (1999) A kinase-regulated PDZ-domain interaction controls endocytic sorting of the beta2-adrenergic receptor. Nature 401: 286-290.[CrossRef][Medline]

Cong M, Perry SJ, Hu LA, Hanson PI, Claing A, and Lefkowitz RJ (2001) Binding of the β2 adrenergic receptor to N-ethylmaleimide-sensitive factor regulates receptor recycling. J Biol Chem 276: 45145-45152.[Abstract/Free Full Text]

Devic E, Xiang Y, Gould D, and Kobilka B (2001) β-Adrenergic receptor subtype-specific signaling in cardiac myocytes from β1 and β2 adrenoceptor knockout mice. Mol Pharmacol 60: 577-583.[Abstract/Free Full Text]

Gage RM, Kim KA, Cao TT, and von Zastrow M (2001) A transplantable sorting signal that is sufficient to mediate rapid recycling of G protein-coupled receptors. J Biol Chem 276: 44712-44720.[Abstract/Free Full Text]

Gage RM, Matveeva EA, Whiteheart SW, and von Zastrow M (2005) Type I PDZ ligands are sufficient to promote rapid recycling of G protein-coupled receptors independent of binding to N-ethylmaleimide-sensitive factor. J Biol Chem 280: 3305-3313.[Abstract/Free Full Text]

Hall RA, Premont RT, Chow CW, Blitzer JT, Pitcher JA, Claing A, Stoffel RH, Barak LS, Shenolikar S, Weinman EJ, et al. (1998) The beta2-adrenergic receptor interacts with the Na+/H+-exchanger regulatory factor to control Na+/H+ exchange. Nature 392: 626-630.[CrossRef][Medline]

Hanley JG, Khatri L, Hanson PI, and Ziff EB (2002) NSF ATPase and alpha-/beta-SNAPs disassemble the AMPA receptor-PICK1 complex. Neuron 34: 53-67.[CrossRef][Medline]

Heydorn A, Sondergaard BP, Ersboll B, Holst B, Nielsen FC, Haft CR, Whistler J, and Schwartz TW (2004) A library of 7TM receptor C-terminal tails. Interactions with the proposed post-endocytic sorting proteins ERM-binding phosphoprotein 50 (EBP50), N-ethylmaleimide-sensitive factor (NSF), sorting nexin 1 (SNX1), and G protein-coupled receptor-associated sorting protein (GASP). J Biol Chem 279: 54291-54303.[Abstract/Free Full Text]

Lu W and Ziff EB (2005) PICK1 interacts with ABP/GRIP to regulate AMPA receptor trafficking. Neuron 47: 407-421.[CrossRef][Medline]

McDonald PH, Cote NL, Lin FT, Premont RT, Pitcher JA, and Lefkowitz RJ (1999) Identification of NSF as a β-arrestin1-binding protein. Implications for β2-adrenergic receptor regulation. J Biol Chem 274: 10677-10680.[Abstract/Free Full Text]

Mialet-Perez J, Green SA, Miller WE, and Liggett SB (2004) A primate-dominant third glycosylation site of the β2-adrenergic receptor routes receptors to degradation during agonist regulation. J Biol Chem 279: 38603-38607.[Abstract/Free Full Text]

Milano CA, Allen LF, Dolber PC, Johnson TD, Rockman HA, Bond RA, and Lefkowitz RJ (1995) Marked enhancement in myocardial function resulting from overexpression of a human beta-adrenergic receptor gene. J Thorac Cardiovasc Surg 109: 236-241.[Abstract/Free Full Text]

Morgan A and Burgoyne RD (2004) Membrane traffic: controlling membrane fusion by modifying NSF. Curr Biol 14: R968-R970.[CrossRef][Medline]

Osten P, Srivastava S, Inman GJ, Vilim FS, Khatri L, Lee LM, States BA, Einheber S, Milner TA, Hanson PI, et al. (1998) The AMPA receptor GluR2 C terminus can mediate a reversible, ATP-dependent interaction with NSF and alpha- and beta-SNAPs. Neuron 21: 99-110.[CrossRef][Medline]

Patterson A, Zhu W, Chow A, Agrawal R, Kosek J, Xiao R, and Kobilka B (2004) Protecting the myocardium: A role for the beta2 adrenergic receptor in the heart. Crit Care Med 32: 1041-1048.[CrossRef][Medline]

Pitcher JA, Freedman NJ, and Lefkowitz RJ (1998) G protein-coupled receptor kinases. Annu Rev Biochem 67: 653-692.[CrossRef][Medline]

Pitcher JA, Payne ES, Csortos C, DePaoli-Roach AA, and Lefkowitz RJ (1995) The G-protein-coupled receptor phosphatase: a protein phosphatase type 2A with a distinct subcellular distribution and substrate specificity. Proc Natl Acad Sci U S A 92: 8343-8347.[Abstract/Free Full Text]

Rochdi MD and Parent JL (2003) Gq-coupled receptor internalization specifically induced by Gq signaling. Regulation by EBP50. J Biol Chem 278: 17827-17837.[Abstract/Free Full Text]

Sibley DR, Strasser RH, Benovic JL, Daniel K, and Lefkowitz RJ (1986) Phosphorylation/dephosphorylation of the β-adrenergic receptor regulates its functional coupling to adenylate cyclase and subcellular distribution. Proc Natl Acad Sci U S A 83: 9408-9412.[Abstract/Free Full Text]

Sneddon WB, Syme CA, Bisello A, Magyar CE, Rochdi MD, Parent JL, Weinman EJ, Abou-Samra AB, and Friedman PA (2003) Activation-independent parathyroid hormone receptor internalization is regulated by NHERF1 (EBP50). J Biol Chem 278: 43787-43796.[Abstract/Free Full Text]

Swaminath G, Xiang Y, Lee TW, Steenhuis J, Parnot C, and Kobilka BK (2004) Sequential binding of agonists to the β2 adrenoceptor. Kinetic evidence for intermediate conformational states. J Biol Chem 279: 686-691.[Abstract/Free Full Text]

Tanowitz M and von Zastrow M (2003) A novel endocytic recycling signal that distinguishes the membrane trafficking of naturally occurring opioid receptors. J Biol Chem 278: 45978-45986.[Abstract/Free Full Text]

von Zastrow M (2003) Mechanisms regulating membrane trafficking of G protein-coupled receptors in the endocytic pathway. Life Sci 74: 217-224.[CrossRef][Medline]

Whiteheart SW and Matveeva EA (2004) Multiple binding proteins suggest diverse functions for the N-ethylmaleimide sensitive factor. J Struct Biol 146: 32-43.[CrossRef][Medline]

Whiteheart SW, Rossnagel K, Buhrow SA, Brunner M, Jaenicke R, and Rothman JE (1994) N-ethylmaleimide-sensitive fusion protein: a trimeric ATPase whose hydrolysis of ATP is required for membrane fusion. J Cell Biol 126: 945-954.[Abstract/Free Full Text]

Xiang Y and Kobilka B (2003) The PDZ-binding motif of the β2-adrenoceptor is essential for physiologic signaling and trafficking in cardiac myocytes. Proc Natl Acad Sci U S A 100: 10776-10781.[Abstract/Free Full Text]

Xiang Y, Naro F, Zoudilova M, Jin SL, Conti M, and Kobilka B (2005) Phosphodiesterase 4D is required for β2 adrenoceptor subtype-specific signaling in cardiac myocytes. Proc Natl Acad Sci U S A 102: 909-914.[Abstract/Free Full Text]

Xiang Y, Rybin VO, Steinberg SF, and Kobilka B (2002) Caveolar localization dictates physiologic signaling of β2-adrenoceptors in neonatal cardiac myocytes. J Biol Chem 277: 34280-34286.[Abstract/Free Full Text]

Zhu WZ, Zheng M, Koch WJ, Lefkowitz RJ, Kobilka BK, and Xiao RP (2001) Dual modulation of cell survival and cell death by β2-adrenergic signaling in adult mouse cardiac myocytes. Proc Natl Acad Sci U S A 98: 1607-1612.[Abstract/Free Full Text]

作者单位：Department of Molecular and Integrative Physiology, University of Illinois at Urbana-Champaign, Urbana, Illinois (Y.W., Y.X.); Program in Pharmaceutical Sciences and Pharmacogenomics and Department of Pharmaceutical Chemistry (B.L.), Department of Psychiatry and Department of Cellular and Molecular

【关键词】  Bradykinin-Induced

    Bradykinin produced at sites of tissue injury and inflammation elicits acute pain and alters the sensitivity of nociceptive neurons to subsequent stimuli. We tested the hypothesis that bradykinin could elicit long-lasting changes in nociceptor function by activating members of the nuclear factor of activated T-cells (NFAT) family of transcription factors. Bradykinin activation of B2 receptors evoked concentration-dependent (EC50 = 6.0 ± 0.3 nM) increases in intracellular Ca2+ concentration ([Ca2+]i) in a proportion of dorsal root ganglion neurons in primary culture. These [Ca2+] increases were sensitive to inhibition of phospholipase C (PLC) and depletion of Ca2+ stores. In neurons expressing a green fluorescent protein (GFP)-NFAT4 fusion protein, a 2-min exposure to bradykinin induced the translocation of GFP-NFAT4 from the cytoplasm to the nucleus. Translocation was partially inhibited by the removal of extracellular Ca2+ and was blocked by inhibition of calcineurin. Furthermore, bradykinin triggered a concentration-dependent increase in NFAT-mediated transcription of a luciferase gene reporter (EC50 = 24.2 ± 0.1 nM). This depended on the B2 receptor, PLC activation, and inositol triphosphate-mediated Ca2+ release. Transcription was not inhibited by capsazepine. Finally, as indicated by quantitative reverse transcription-polymerase chain reaction, bradykinin elicited an increase in cyclooxygenase mRNA. This increase was sensitive to calcineurin and B2 receptor inhibition. These findings suggest a mechanism by which short-lived bradykinin-mediated stimuli can enact lasting changes in nociceptor function and sensitivity.

    Tissue damage and inflammation result in the production and release of numerous algesic and proinflammatory agents that act to elicit pain or lower the threshold of peripheral nociceptive neurons to painful stimuli. One of these agents, the nonapeptide bradykinin, directly evokes pain, decreases the activation threshold of sensory neurons, and elicits many of the hallmark signs of inflammation, including edema, redness, and local heat (Dray and Perkins, 1993; Marceau and Regoli, 2004).

    Bradykinin is produced at the site of tissue injury via cleavage of a kininogen precursor by the protease kallikrein (Dray and Perkins, 1993). Two bradykinin receptors have been identified: a constitutive B2 receptor, and an inducible B1 receptor. Both receptors are present on the peripheral termini of sensory nerves (Steranka et al., 1988). These receptors couple through heterotrimeric G-proteins to activate phospholipase A2 and phospholipase C (PLC). PLC stimulation results in the generation of diacylglycerol and inositol triphosphate (IP)3 and release of Ca2+ from the endoplasmic reticulum (ER) (Thayer et al., 1988b). Activation of phospholipase A2 liberates arachidonic acid leading to the production of prostaglandins by cyclooxygenases and synthesis of 12-hydroperoxy-eicosatetraenoic acid via 12-lipoxygenase (Shin et al., 2002). Downstream targets of these signaling cascades contribute to changes in nociceptor sensitivity.

    Bradykinin elicits rapid changes in the response characteristics of sensory neurons. Activation of B2 receptors decreases the threshold for nociceptor activation by heat and other stimuli by sensitizing transient receptor potential vanilloid receptor (TRPV1) channels (Premkumar and Ahern, 2000; Chuang et al., 2001). It also inhibits K+ conductances in afferent neurons, increasing neuronal excitability (Usachev et al., 2002; Oh and Weinreich, 2004). Bradykinin produces pain hypersensitivity by potentiating glutamatergic transmission between primary afferents and dorsal horn neurons (Wang et al., 2005).

    Tissue damage and inflammation also produce delayed changes in nociceptor function by inducing transcriptional changes. Increased expression of numerous genes, including Substance P, calcitonin gene-related peptide, brain-derived neurotrophic factor (BDNF), growth-associated protein-43, and ion channels Nav1.8, TRPV1, and acid-sensing ion channel, are seen in models of inflammation and nerve damage. Furthermore, inflammatory hyperalgesia is mediated in part by phenotypic switches in the sensory modality of DRG neurons that require changes in gene expression (Woolf and Costigan, 1999).

    We were interested in the idea that bradykinin, by regulating [Ca2+]i, stimulates changes in gene expression that may underlie long-term changes in the response characteristics of sensory neurons. We hypothesized that bradykinin-induced Ca2+ increases would activate members of the NFAT family of transcription factors.

    NFAT activation of gene transcription is regulated by Ca2+ (Dolmetsch et al., 1998). In unstimulated cells, NFAT is phosphorylated and restricted to the cytoplasm. After an increase in [Ca2+]i, cytoplasmic NFAT is dephosphorylated by the Ca2+-dependent serine/threonine phosphatase calcineurin (Beals et al., 1997). This exposes a nuclear localization signal, allowing NFAT to translocate to the nucleus and initiate transcription through interaction with other transcription factors. Four isoforms of NFAT (NFAT1-4) transcription factors have been identified, several of which have been localized to neuronal tissue (Graef et al., 2003; Groth and Mermelstein, 2003), including DRG neurons (Kim et al., 2006). Increases in [Ca2+]i and activation of NFAT-mediated transcription have been implicated in the maintenance of pain in response to Substance P in spinal neurons (Seybold et al., 2006).

    Here we demonstrate that bradykinin-evoked Ca2+ release from the ER of rat DRG neurons initiated the translocation of NFAT to the nucleus and subsequent activation of NFAT-dependent transcription. We also demonstrate that bradykinin increases the level of mRNA for the proinflammatory enzyme Cox-2 in a calcineurin-dependent manner. These data suggest that in sensory neurons, NFAT conveys proinflammatory signals to the nucleus to initiate long-term changes in sensory neuron function.

    Cell Culture. Rat DRG neurons were grown in primary culture as described previously (Thayer et al., 1988b). In brief, 1- to 3-day-old Sprague-Dawley rats were killed by decapitation with sharp scissors under a protocol approved by the University of Minnesota Institutional Animal Care and Use Committee. Ganglia were dissected from the thoracic and lumbar regions, incubated at 37°C in collagenase-dispase (Vibrio alginolyticus/Bacillus polymyxa, 0.8 and 6.4 U/ml, respectively; Roche Diagnostics, Indianapolis, IN) for 45 min and dissociated by trituration through a flame-constricted Pasteur pipette. Non-neuronal cells attach more readily to substrate than do neuronal cells (Seybold et al., 2006). We plated aliquots of dissociated cells onto HNO3-washed glass coverslips for 1 h and then replated the unattached neurons onto laminin-coated (50 mg/ml) glass coverslips (25 mm diameter). Cells were grown in Ham's F-12 medium supplemented with 5% heat-inactivated horse serum, 5% fetal bovine serum, 50 ng/ml NGF-7S (mouse submaxillary gland; Sigma, St. Louis, MO), 4.4 mM glucose, 2 mM L-glutamine, modified Eagle's medium vitamins, and penicillin/streptomycin (100 U/ml and 100 mg/ml, respectively). Cultures were maintained at 37°C in a humidified atmosphere of 5% CO2. Because neurotrophins activate NFAT-dependent transcription, serum and NGF were replaced with 1% B27 supplement (Invitrogen, Carlsbad, CA) 24 h before imaging experiments to minimize background NFAT activity (Groth and Mermelstein, 2003). Cells were used on the third and fourth day in vitro.

    [Ca2+] Measurement. [Ca2+]i was determined with the Ca2+-sensitive fluorescent dye fura-2 (Grynkiewicz et al., 1985). Cells were loaded with indicator by incubation with 5 µM fura-2 acetoxymethyl ester for 45 min at 37°C in HEPES-buffered Hanks' salt solution (HHSS), pH 7.45, containing 0.5% bovine serum albumin. HHSS was composed of the following: 20 mM HEPES, 137 mM NaCl; 1.3 mM CaCl2, 0.4 mM MgSO4, 0.5 mM MgCl2, 5.4 mM KCl, 0.4 mM KH2PO4, 0.3 mM Na2HPO4, 3.0 mM NaHCO3, and 5.6 mM glucose. Coverslips with loaded cells were mounted in a flow-through chamber for viewing (Thayer et al., 1988b) (10-s solution exchange) that was placed on the stage of an inverted Olympus IX70 microscope (Olympus Optical, Tokyo, Japan) equipped with a 40x objective (UApo/340, numerical aperture = 1.35). Cells were superfused with HHSS at a rate of 1.0 to 1.5 ml/min for 10 min before starting an experiment. Fura-2-based digital imaging was performed using an Optoscan monochromator (Cairn Research LTD, Faversham, Kent, UK) rapidly switching excitation between 340 nm (8 nm slit width) and 380 (8) nm. Emission was detected at 517 (30) nm with a Cascade cooled charge-coupled device camera (Roper, Tucson, AZ). Images were acquired and analyzed using Metafluor software (Molecular Devices, Sunnyvale CA).

    Fluorescence changes were converted to [Ca2+]i by using the formula [Ca2+]i = Kd β (R-Rmin)/(Rmax-R), where R is 340/380 nm fluorescence ratio (Grynkiewicz et al., 1985). The dissociation constant (Kd) for fura-2 was 140 nM and β was the ratio of fluorescence emitted at 380 nm measured in the absence and presence of Ca2+. Rmin, Rmax, and β were determined by bathing intact cells in 2 µM ionomycin in Ca2+-free buffer (1 mM EGTA) and saturating Ca2+ (5 mM Ca2+). Values for Rmin, Rmax, and β were 0.237, 4.10, and 6.52, respectively.

    Transfection. Gene transfer into DRG neurons was performed as described previously (Usachev et al., 2000). In brief, plasmid DNA was precipitated onto gold particles (1.6 µm) and introduced into DRG neurons using a Biolistic particle delivery system (PDS-1000; Bio-Rad Laboratories, Hercules, CA).

    Dual-Luciferase-Based Gene Reporter Assays. DRG neurons were transferred to serum-free Ham's F-12 medium supplemented with 1% B27 (Invitrogen) 24 h after plating. Five hours after serum removal, DRG neurons were cotransfected with plasmids encoding a luciferase (firefly) based reporter of NFAT activity (pNFAT-luciferase; Graef et al., 1999) and a constitutively active Renilla reniformis luciferase under the control of a thymidine kinase promoter (pRL-TK; Promega, Madison, WI) at a ratio of 5:1. The neurons were stimulated 2 h after transfection and returned to culture medium for an additional 16 to 18 h. Neurons from a single coverslip were lysed and the activity of each reporter measured on a TD 20/20 luminometer (Turner Biosystems, Sunnyvale, CA) using the Dual-Luciferase Reporter Assay (Promega). The expression of firefly luciferase was normalized to constitutively expressed R. reniformis luciferase activity to correct for differences in transfection efficiency. Within each experiment, treatments were conducted in duplicate to triplicate. Each experiment was conducted at least three times on cultures prepared from separate litters.

    Simultaneous Confocal Imaging of [Ca2+]i and GFP-NFAT4. DRG cultures were transfected with a plasmid encoding GFP-NFAT4 fusion protein (Tomida et al., 2003). Neurons were transferred to serum-free Ham's F-12 medium supplemented with 1% B27 (Invitrogen) 24 h after plating. Forty-eight hours after transfection, cultures were loaded with X-Rhod-1 acetoxymethyl ester (2 µM at room temperature for 45 min). Cells were rinsed in HHSS, and the indicator was allowed to de-esterify for 15 min before the start of the experiment. GFP and [Ca2+] imaging were performed on a Fluoview 300 laser scanning confocal microscope attached to an inverted microscope (Olympus IX70) equipped with a PlanApo 60x objective (numerical aperture = 1.40) (Olympus Optical). GFP-NFAT4 and X-Rhod-1 were excited with the 488 nm (Argon) and 540 nm (HeNe) laser lines, and the fluorescence was imaged at 510 to 540 nm and >605 nm, respectively (Jackson and Thayer, 2006).

    Fig. 1. Bradykinin-elicited [Ca2+]i increases in rat DRG neurons in culture. [Ca2+]i was measured in a field of DRG neurons in culture using fura-2-based digital [Ca2+] imaging as described under Materials and Methods. Drugs were applied by superfusion at the times indicated by the horizontal bars. A, bradykinin elicited a concentration-dependent increase in [Ca2+]i in rat DRG neurons. Data points represent at least three experiments and are expressed as mean ± S.E.M. Curves were fitted by a logistic equation of the form [Ca2+] = Ca2+max/(1 + 10^(logEC50-X)), where X is the logarithm of the bradykinin concentration, Ca2+max is the baseline-corrected peak [Ca2+], and the Hill coefficient is 1. B, bradykinin (BK; 10 nM, 2-min application) elicited [Ca2+]i increases of reproducible amplitude. Bradykinin was applied at 20-min intervals as indicated. C, the bradykinin-induced [Ca2+]i increase was blocked by the B2-selective antagonist HOE140 (1 µM). D, bar chart depicts the effect of inhibitors using the testing paradigm depicted in C. Concentrations of inhibitors are as follows: HOE140, 1 µM (n = 9); Cyclosporin A, 1 µM (n = 14); U-73122, 1 µM (n = 19); U-73343,1 µM (n = 15); 0 Ca2+ (Ca2+-free medium + 20 µM EGTA; n = 46 neurons); capsazepine, 1 µM (n = 30); ryanodine, 1 µM (n = 19); and cyclopiazonic acid, 5 µM (n = 66). Each experiment was performed on at least three fields of cells; n values represent the number of cells. Data are presented as mean ± S.E.M., and significance was determined by analysis of variance using Dunnett's post hoc test. **, p < 0.01 relative to control.

    Quantitative Real-Time PCR. DRG neurons were prepared using the preplating protocol. Medium was replaced with serum-free Ham's F-12 medium supplemented with 1% B27 2 h after plating. On day 2 in vitro, growth medium was removed, and neurons were stimulated by application of bradykinin (1 µM, 2 min). Stimulation was terminated by removal of bradykinin and return of growth medium. Total RNA was isolated from DRG neurons using an RNeasy Mini Kit (QIAGEN, Valencia, CA) 7 h after bradykinin stimulation and further purified by DNase digestion. RNA was reverse transcribed into cDNA using a QuantiTect RT-PCR kit (QIAGEN). Real-time PCR was performed using Sybr Green Master Mix (Applied Biosystems, Foster City, CA) on an Applied Biosystems 7300 Real-Time PCR System. Samples were run through 40 cycles (95°C for 10 s, 58°C for 33 s, 72°C for 33 s). Each cDNA sample was run in duplicate for the target (e.g., Cox-2) and the normalizing gene (S15). Experiments were repeated at least three times using RNA collected from at least three separate neuronal platings. Amplicon specificity was determined by melting curve analysis and gel electrophoresis. Primer pair sequences were validated previously (Groth et al., 2007) and are as follows: BDNF (GenBank accession number NM_007540[GenBank]): forward primer, 5'-CCA TAA AGG ACG CGG ACT TGT ACA-3'; reverse primer, 3'-AGA CAT GTT TGC GGC ATC CAG-3'; Cox-2 (S67722[GenBank]): forward primer, 5'-GCT GCT GCC GGA CAC CTT CA-3'; reverse primer, 5'-AGC AAC CCG GCC AGC AAT CT-3'; and S15 (BC094409[GenBank]): forward primer, 5'-CCG AAG TGG AGC AGA AGA AG-3'; reverse primer 5'-CTC CAC CTG GTT GAA GGT C-3'. All primers were synthesized by Integrated DNA Technologies (Coralville, IA). Cycle thresholds (Ct) were calculated automatically using the Applied Biosystems software to minimize user bias. -Fold increases in mRNA expression for each sample were calculated using the Pfaffl correction of the Livak method in which Ct values of target genes are normalized to that of S15 (Pfaffl, 2001) using the following equation: Fold increase = (Etarget)Ct target (control-treated)/(Etarget)Ct S15(control-treated), where E is the efficiency of the primer. To determine individual primer efficiencies, we serially diluted cDNA samples and calculated Ct values for each dilution. Ct values were plotted against the log [cDNA] and the efficiency calculated from the formula E = 10(-1/slope).

    Bradykinin Activates B2 Receptors to Mobilize IP3-Sensitive Ca2+ Stores in DRG Neurons. We monitored changes in [Ca2+]i in a field of DRG neurons grown in primary culture using fura-2 based digital [Ca2+]i imaging. The resting [Ca2+]i before bradykinin exposure was 71 ± 1nM (n = 1176 neurons). Application of bradykinin (2 min) by superfusion elicited concentration-dependent increases in [Ca2+]i (Fig. 1A). Because only a subset of DRG neurons expresses receptors for bradykinin (Thayer et al., 1988a), we separated nonresponders from weak responders by maximally stimulating the cells with 1 µM bradykinin 20 min after the application of the test concentration. Neurons that failed to display a net increase in [Ca2+]i of at least 25 nM in response to this second application of bradykinin were considered nonresponders and were not included in further analysis. Fitting of the bradykinin concentration-response data with a logistic equation determined a 50% effective concentration (EC50) of 6.0 ± 0.3 nM. This concentration-response profile is similar to those seen in DRG neurons and other cell types (Lo and Thayer, 1993).

    Bradykinin (10 nM, 2 min) evoked reproducible increases in [Ca2+]i from the same DRG neuron (Fig. 1B). The amplitude of the first [Ca2+]i response to 10 nM bradykinin was 84 ± 5 nM. The amplitude of the second bradykinin-evoked response (peak 2) was 86 ± 6% (n = 169 neurons) of the first [Ca2+]i transient (peak 1), indicating that a 2-min exposure to 10 nM bradykinin did not appreciably desensitize its receptor and that sufficient time elapsed to allow refilling of the Ca2+ stores. Because individual DRG neurons vary greatly in the amplitude of their bradykinin-induced Ca2+ responses, we normalized the amplitude of the second response to that of the first response (peak 2/peak 1). We used this as an assay to test the regulation of bradykinin-induced Ca2+ signals.

    Bradykinin acts through either of two G-protein coupled receptors (B1 or B2) present in sensory neurons. The response to bradykinin was abolished by application of the B2 receptor antagonist HOE140 (1 µM, n = 9 neurons), indicating that B1 receptor activation is unlikely to contribute to the responses we observed (Fig. 1C). B2 receptors couple to several signaling cascades. To assess the contribution of extracellular Ca2+ to the bradykinin response, we applied Ca2+-free media (20 µM EGTA) to the cells 1 min before and throughout application of bradykinin. Removal of extracellular Ca2+ decreased the amplitude of the [Ca2+]i response by 35% relative to control (n = 46 neurons, p < 0.01), suggesting that intracellular Ca2+ release was sufficient to account for much of the bradykinin-induced Ca2+ signal (Fig. 1D). Because removal of extracellular [Ca2+] only partially inhibited the Ca2+ signal, we next tested the contribution of the intracellular Ca2+ stores to the bradykinin-induced [Ca2+]i increase using the sarcoplasmic-endoplasmic reticulum Ca2+-ATPase inhibitor cyclopiazonic acid (CPA; 5 µM). Inhibition of ER Ca2+ pumps depleted the ER Ca2+ stores and decreased the amplitude of [Ca2+]i transients evoked by bradykinin by 56% (n = 66 neurons, p < 0.01).

    Fig. 2. Bradykinin triggers translocation of GFP-NFAT4 coincident with [Ca2+]i increases. [Ca2+]i and GFP-NFAT4 were simultaneously imaged in single, transfected DRG neurons in culture using the argon and HeNe laser lines of a confocal microscope to excite the GFP-NFAT4 and the Ca2+ indicator X-Rhod-1. Drugs were applied by superfusion at the times indicated by the horizontal bars. A, representative images depict translocation of GFP-NFAT4 from the cytosol to the nucleus. GFP-NFAT4 was localized primarily to the cytosol in unstimulated neurons. Images were acquired at times indicated in B. B, application of 1 µM bradykinin (BK; 2 min) elicited a transient increase in [Ca2+]i (red) and increased the proportion of GFP-NFAT4 residing in the nucleus [NFAT (nuclear/cytoplasmic); black]. The proportion of GFP-NFAT4 residing in the nucleus was quantified as a ratio of the mean GFP fluorescence from the nucleus relative to the cytoplasm. Changes in X-Rhod-1 (red) fluorescence (F) were normalized to its initial fluorescence intensity (F0) from a region encompassing the DRG cell body. C, depolarization (90 mM K+; 2 min) elicited an increase in [Ca2+]i (red) and increased the proportion of GFP-NFAT4 residing in the nucleus [NFAT (nuclear/cytoplasmic); black].

    Bradykinin is known to evoke IP3-mediated Ca2+ release in neurons. To verify that this signaling pathway was required for bradykinin-induced Ca2+ responses, we treated DRG neurons with the phospholipase C inhibitor U-73122. U-73122 decreased the amplitude of the second Ca2+ transient by 86% (n = 19; p < 0.01). However, this response was not inhibited by its inactive analog U-73343 (1 µM; peak 2/peak 1 = 0.78 ± 0.13; n = 15), demonstrating the specificity of the U-73122-induced block. DRG neurons also possess ryanodine-sensitive Ca2+ stores (Shmigol et al., 1995). We tested the contribution of these stores to the bradykinin-induced [Ca2+] transients by application of ryanodine. Ryanodine (1 µM) failed to inhibit the response (peak 2/peak 1 = 0.91 ± 0.08; n = 36).

    The immunosuppressant cyclosporine A interacts with multiple proteins involved in Ca2+ signaling, including calcineurin, immunophilins, and the mitochondrial permeability transition pore (Snyder et al., 1998). These interactions can exert complex effects on Ca2+ signals. However, bradykinin-evoked [Ca2+]i transients were not affected by application of cyclosporin A (10 µM; peak 2/peak 1 = 0.89 ± 0.17; n = 14).

    Bradykinin decreases the temperature threshold for vanilloid receptor activation and increases the membrane current activated by heat (Cesare et al., 1999). Activation of protein kinase C by bradykinin or phorbol esters, although controversial (Bhave et al., 2003), has been proposed to directly evoke TRPV1-dependent currents and Ca2+ increases (Premkumar and Ahern, 2000), an effect suppressed by the competitive TRPV1 antagonist capsazepine. We tested whether [Ca2+]i transients evoked by bradykinin were caused by influx through TRPV1 channels by applying the TRPV1-competitive antagonist capsazepine. Capsazepine (1 µM) did not affect the [Ca2+]i response (peak 2/peak 1 = 0.79 ± 0.1, n = 30), suggesting that TRPV1-mediated Ca2+ influx did not contribute to the bradykinin-evoked [Ca2+]i increase.

    Bradykinin Triggers Translocation of NFAT. In addition to generating acute pain, bradykinin elicits long-lasting changes in the sensitivity of nociceptors to subsequent painful stimuli (Woolf and Costigan, 1999). Here we tested the hypothesis that bradykinin-evoked [Ca2+]i increases triggered the activation and translocation of the transcription factor NFAT. DRG neurons were transfected with a plasmid encoding enhanced green fluorescent protein fused to the N terminus of NFAT4 (GFP-NFAT4) using a gene gun (Tomida et al., 2003). Previous work demonstrated that this method preferentially transfects sensory neurons without impairing physiological function (Usachev et al., 2000).

    NFAT localization and activity is controlled by its phosphorylation state (Crabtree and Olson, 2002; Groth and Mermelstein, 2003). NFAT is dephosphorylated by the phosphatase calcineurin and subsequently translocates to the nucleus where it interacts with other transcription factors to initiate transcription. We quantified NFAT translocation by measuring the mean GFP fluorescence from regions in the cytoplasm and nucleus and calculating the nuclear-to-cytosolic ratio [GFP-NFAT4(n/c)]. [Ca2+]i was monitored simultaneously using the fluorescent Ca2+ indicator X-Rhod-1 acetoxymethyl ester (Kd = 700 nM). Resting GFP-NFAT4(n/c) was 0.29 ± 0.01 (n = 116 neurons). Application of bradykinin (1 µM) elicited an increase in [Ca2+]i (F/F0 = 0.8 ± 0.1, n = 17) and a corresponding translocation of GFP-NFAT4 into the nucleus [GFP-NFAT4(n/c) = 0.9 ± 0.2] (Fig. 2, A and B). The [Ca2+]i elevation was rapid and transient, comparable in wave form with that recorded with fura-2 (Fig. 1B), whereas the translocation developed slowly, reaching a maxima over the course of 20 to 30 min. GFP-NFAT4 continued to accumulate in the nucleus even after [Ca2+]i had recovered to basal levels. Thus, a transient bradykinin-elicited [Ca2+]i increase can trigger long-lasting changes in cell signaling. Depolarization-induced activation of voltage-gated Ca2+ channels is a powerful activator of NFAT in hippocampal neurons (Graef et al., 1999). Depolarization of DRG neurons with 90 mM K+ evoked a large increase in [Ca2+]i (F/F0 = 1.58 ± 0.21; n = 21) that elicited robust translocation of GFP-NFAT4 [GFP-NFAT4(n/c) = 1.4 ± 0.2; Fig. 2C]. Pretreatment with the calcineurin inhibitor CSA (10 µM, 10 min) blocked NFAT translocation [GFP-NFAT4(n/c) = 0.22 ± 0.06; n = 4] but did not seem to inhibit the bradykinin-induced [Ca2+]i increase (F/F0 = 0.85 ± 0.15; Fig. 3A and Fig. 1D), consistent with translocation requiring NFAT dephosphorylation. GFP-NFAT4 translocation was triggered by Ca2+ release from the ER evoked by 5 µM CPA [GFP-NFAT4(n/c) = 2.4 ± 0.8; F/F0 = 0.9 ± 0.35; n = 4; Fig. 3B]. To verify that the bradykinin-mediated NFAT translocation was dependent on Ca2+ release, we applied bradykinin (1 µM) in Ca2+-free solution (+20 µM EGTA). In the absence of extracellular Ca2+, bradykinin still elicited a [Ca2+]i increase (F/F0 = 0.73 ± 0.22; n = 4) and a corresponding translocation of NFAT to the nucleus [GFP-NFAT4(n/c) = 0.4 ± 0.1; Fig. 3C], although to a lesser degree than in normal Ca2+ buffer (Fig. 2B). These data suggest that Ca2+ mobilization from the ER and Ca2+ influx contribute to the bradykinin-induced translocation of NFAT.

    Fig. 3. Bradykinin-induced GFP-NFAT4 translocation depends on Ca2+ release and calcineurin activation. [Ca2+]i and GFP-NFAT4 were simultaneously imaged in single, transfected DRG neurons in culture using the argon and HeNe laser lines of a confocal microscope to excite the GFP-NFAT4 and the Ca2+ indicator X-Rhod-1. Drugs were applied by superfusion at the times indicated by the horizontal bars. A, representative trace depicts the effects of calcineurin inhibition (1 µM CSA) on GFP-NFAT4 translocation. CSA inhibited GFP-NFAT4 translocation without altering the amplitude of the [Ca2+]i transient (n = 4). One of four bradykinin-induced [Ca2+] responses in CSA exhibited oscillations, a response also seen in untreated cells (n = 2) that showed robust GFP-NFAT4 translocation [GFP-NFAT4(n/c) = 1.1]. B, representative trace depicts the effect of store mobilization on GFP-NFAT4 localization. Blocking sarcoplasmic-endoplasmic reticulum Ca2+-ATPase with CPA (5 µM, 30 min) elicited a [Ca2+]i increase that caused GFP-NFAT4 translocation (n = 4). C, Ca2+-mediated GFP-NFAT4 translocation occurs in the absence of Ca2+ influx. Representative trace shows bradykinin-evoked (1 µM) [Ca2+]i increases and GFP-NFAT4 translocation in Ca2+-free medium (20 µM EGTA, n = 4).

    Bradykinin Evokes NFAT-Dependent Transcription. Bradykinin-induced translocation of GFP-NFAT4 indicates that bradykinin activates NFAT-dependent transcription. To determine whether bradykinin activates endogenous NFAT in DRG neurons, we transfected DRG cultures with an NFAT-luciferase (firefly) expression reporter (Graef et al., 1999). Firefly luciferase activity (relative light units; RLUFirefly) was normalized to cotransfected, constitutively expressed R. reniformis luciferase activity (RLURenilla). In unstimulated neurons, the resting ratio was 0.22 ± 0.02 (n = 15). Stimulation with bradykinin caused a concentration-dependent increase in NFAT-dependent transcription. Fitting of the concentration-response data with a logistic equation yielded an EC50 of 24.2 ± 0.1 nM (Fig. 4A).A1 µM concentration of bradykinin maximally stimulated NFAT-dependent transcription (RLUFirefly/RLURenilla = 0.40 ± 0.04; n = 15; p < 0.05; Fig. 4B). Values were comparable with those evoked by a 2-min depolarization in 90 mM K+ (RLUFirefly/RLURenilla = 0.38 ± 0.03; n = 15; p < 0.05). Mobilization of Ca2+ stores with CPA produced a smaller increase in NFAT-mediated transcription (RLUFirefly/RLURenilla = 0.29 ± 0.08; n = 3). As shown in Fig. 4C, pretreatment with CSA (10 µM, 15 min) abolished bradykinin-induced NFAT-dependent transcription (0.20 ± 0.02; p < 0.05; n = 6), consistent with the inhibition of NFAT translocation by this drug (Fig. 3A). This effect was not caused by changes in IP3-mediated [Ca2+]i increases because CSA failed to alter the amplitude of bradykinin-induced [Ca2+]i transients (Figs. 1D and 3A).

    Fig. 4. Bradykinin stimulates NFAT-dependent transcription. Luciferase activity, expressed as the ratio of firefly to R. reniformis luminescence, was measured in a population of transfected DRG neurons as described under Materials and Methods. Data are presented as mean ± S.E.M. Significance was determined by Student's t test. A, bradykinin elicited concentration-dependent increases in NFAT-dependent transcription. Data points represent four separate experiments from different neuronal platings for which full concentration-responses were run in parallel. Curves were fitted by a logistic equation of the form RLUFirefly/RLURenilla = Rmin + [(Rmax-Rmin)/(1 + 10^(LogEC50-X)], where X is the logarithm of the bradykinin concentration, Rmin is the RLUFirefly/RLURenilla in unstimulated cells, Rmax is RLUFirefly/RLURenilla after 1 µM bradykinin, and the Hill coefficient is 1. *, p < 0.05 versus control. B, activation of NFAT-dependent transcription is not Ca2+ source-specific. Bradykinin (1 µM BK; 2 min, n = 15), depolarization (90 mM K+; 2 min, n = 15), and CPA (5 µM; 2 min, n = 3) all elicit NFAT-dependent transcription. *, p < 0.05 versus control. C, bradykinin-induced increases in NFAT-dependent transcription require activation of B2 receptors, calcineurin, and IP3-mediated Ca2+ release. Coverslips were preincubated (15 min) with either HOE140 (1 µM, n = 3), U-73122 (1 µM, n = 9), U-73343 (1 µM, n = 9), CSA (1 µM, n = 6), capsazepine (CZP, 1 µM; n = 3), ryanodine (1 µM, n = 5), or xestospongin C (xest c; 500 nM, n = 5) before being challenged with bradykinin 100 nM. *, p < 0.05 versus control; #, p < 0.05 versus bradykinin.

    Bradykinin may act through several different pathways. To further elucidate the mechanism of bradykinin-induced NFAT-dependent transcription, we tested several of these possibilities. Stimulation of NFAT-dependent transcription by bradykinin was inhibited by the selective B2 receptor antagonist HOE140 (1 µM; RLUFirefly/RLURenilla = 0.21 ± 0.02; p < 0.05; n = 3; Fig. 4C), consistent with the complete block of the bradykinin-induced [Ca2+]i increase produced by this drug (Fig. 1, C and D). Bradykinin-elicited transcription was blocked by the PLC antagonist U-73122 (1 µM; RLUFirefly/RLURenilla = 0.31 ± 0.03; n = 9; p < 0.05). The specificity of this inhibition was demonstrated by the failure of its inactive analog U-73343 (1 µM; RLUFirefly/RLURenilla = 0.43 ± 0.05; n = 9) to affect luciferase expression. Although these data imply that IP3-mediated [Ca2+]i release was necessary, we tested this explicitly using the IP3 receptor antagonist xestospongin C. Pretreatment (15 min) with xestospongin C (500 nM) completely inhibited the bradykinin-mediated increase in luciferase activity (500 nM; RLUFirefly/RLURenilla = 0.21 ± 0.03; n = 5; p < 0.05). This inhibition was not mimicked by blockade of caffeine-sensitive Ca2+ stores by application of ryanodine (1 µM; RLUFirefly/RLURenilla = 0.45 ± 0.1; n = 5). Bradykinin sensitizes the TRPV1 receptor to heat (Chuang et al., 2001). However, bradykinin-mediated increases in NFAT-dependent transcription were not antagonized by pretreatment (1 µM, 15 min) with capsazepine (RLUFirefly/RLURenilla = 0.43 ± 0.1; n = 3). Bradykinin Increases Transcription of Cox-2 mRNA.

    Bradykinin is produced at sites of tissue injury (Barlas et al., 1985), and increases in bradykinin concentration have been noted at sites of inflammation (Hargreaves et al., 1988). We tested whether bradykinin activation of NFAT increased the expression of BDNF and Cox-2 using quantitative real-time PCR. Both genes have NFAT binding sites in their promoter regions (Iñiguez et al., 2000; Groth and Mermelstein, 2003), and both proteins are implicated in inflammatory pain (Woolf and Costigan, 1999; Svensson and Yaksh, 2002). Bradykinin (1 µM, 2 min) was applied to cultured rat DRG neurons and RNA harvested 7 h after stimulation. Bradykinin produced a 2.0 ± 0.43 (n = 5)-fold increase in Cox-2 mRNA relative to sham-treated controls. To determine whether this response depended on calcineurin activation, we pretreated neurons with CSA (10 µM) 15 min before and during bradykinin exposure. RNA was harvested 7 h after stimulation with bradykinin. Pretreatment with CSA not only blocked the bradykinin-mediated increase in Cox-2 mRNA but reduced Cox-2 mRNA to below control levels (0.51 ± 0.23, n = 3), suggesting tonic NFAT-mediated transcription of Cox-2. Bradykinin-induced Cox-2 mRNA expression was attenuated by pretreatment with HOE140 (1 µM, 1.18 ± 0.19, n = 3), suggesting that it depends on the selective activation of the bradykinin B2 receptor. Activation of NFAT increases the transcription of BDNF in hippocampal neurons (Groth and Mermelstein, 2003). However, we did not detect any significant changes in BDNF mRNA (0.7 ± 0.3-fold change, n = 3) in DRG neurons when challenged with bradykinin.

    Application or injection of bradykinin directly elicits pain and sensitizes DRG neurons to painful stimuli (Dray and Perkins, 1993). We describe for the first time a pathway linking bradykinin-induced mobilization of Ca2+ stores to translocation of NFAT, initiation of transcription, and subsequent increases in mRNA for the proinflammatory enzyme cyclooxygenase-2. This mechanism provides a pathway by which bradykinin may contribute to long-lasting changes in the sensitivity of sensory neurons to painful stimuli.

    Bradykinin-Induced Regulation of Ca2+ Release and NFAT Activation. A subset of DRG neurons expresses receptors for bradykinin (Thayer et al., 1988a; Cesare et al., 1999). These receptors are preferentially found on small-sized DRG neurons involved in nociception (Dray and Perkins, 1993), suggesting that the gene expression studies described in Figs. 4 and 5 probably underestimate the magnitude of the bradykinin-evoked responses because only approximately 30% of the DRG neurons express the appropriate receptor. Bradykinin stimulation increases [Ca2+]i via Ca2+ release from the ER and influx across the plasma membrane. Our results build on the model established previously in which bradykinin activation of B2 receptors triggers the cleavage of phosphatidyl 4,5-bisphosphate by PLC to diacylglycerol and IP3 with the subsequent release of Ca2+ from intracellular stores (Thayer et al., 1988a; Dray and Perkins, 1993). In our neuronal cultures, we showed that bradykinin was capable of eliciting Ca2+ release from the ER in a manner that was attenuated by PLC inhibition with U-73122 and by depletion of the ER with CPA, suggesting that bradykinin triggered Ca2+ release from intracellular stores. That neither of these treatments completely blocked bradykinin-induced [Ca2+]i transients suggests that influx also contributed to these increases. Furthermore, although our results demonstrate that IP3-mediated Ca2+ release is sufficient to trigger translocation of NFAT to the nucleus, the extent of translocation in Ca2+-free solution was decreased relative to control (Fig. 3B). In contrast, bradykinin-induced NFAT-mediated transcription was completely blocked by xestospongin C, suggesting that Ca2+ store mobilization is necessary for bradykinin-triggered NFAT-dependent transcription. Perhaps xestospongin inhibits capacitative calcium entry caused by blocking IP3-mediated Ca2+ release. There is a precedent in neurons for the activation of NFAT by other mechanisms. Ca2+ entry via L-type Ca2+ channels (Graef et al., 2003) and IP3-mediated Ca2+ release (Groth and Mermelstein, 2003) activate NFAT3 in hippocampal neurons, suggesting that NFAT signaling can be initiated through multiple mechanisms linked to increases in [Ca2+]i.

    Fig. 5. Bradykinin stimulates increased Cox-2 expression. Quantitative, real-time PCR experiments were performed as described under Materials and Methods. Bradykinin (1 µM, 2 min) stimulation of cultured DRG neurons increased Cox-2 mRNA (n = 5). This effect was attenuated by pretreatment with CSA (10 µM, 15 min, n = 3) and HOE140 (1 µM, 15 min, n = 3). Changes in gene expression are reported as a ratio of fold change relative to untreated control, both normalized to an internal standard (S15). Data are presented as mean ± S.E.M., and significance was determined by Student's t test. *, p < 0.05 versus control; #, p < 0.05 versus bradykinin.

    The addition of growth factors stimulates the activation of NFAT in hippocampal neurons (Groth and Mermelstein, 2003). DRG cultures were maintained in serum-free medium without growth factors to avoid masking changes in NFAT activation after bradykinin challenge. These conditions probably dampened the responsiveness of DRG neurons to bradykinin because expression of B2 receptors may be regulated by NGF (Lee et al., 2002b).

    NFAT Integrates Proinflammatory Signals. Although short-term changes in neuronal activity may be accomplished through modulation of pre-existing proteins (e.g., phosphorylation, glycosylation), long-term changes require transcription and additional protein synthesis. NFAT(1-4) proteins are poised to transduce electrical, growth factor, and inflammatory signals to regulate synaptogenesis (Yoshida and Mishina, 2005), axonal outgrowth (Graef et al., 2003), and survival (Benedito et al., 2005). Here, we demonstrate that bradykinin drives translocation of NFAT4, a major NFAT isoform, in DRG neurons (Kim et al., 2006). We further demonstrate that bradykinin is capable of stimulating NFAT-dependent transcription. In hippocampal and spinal neurons, the growth factors NGF and BDNF initiate NFAT-dependent transcription (Groth and Mermelstein, 2003; Seybold et al., 2006). In spinal neurons, the algesic agent Substance P triggers NFAT-dependent transcription. These results suggest that a number of algesic and proinflammatory agents initiate signaling cascades that converge on the NFAT transcription factors. This in turn suggests a role for NFAT in integrating responses to pain and inflammation in sensory neurons as shown here and in spinal neurons (Seybold et al., 2006).

    Given that all of these growth and inflammatory signals converge on NFAT activation, it remains to be seen to what extent these signals summate. Our data demonstrate that increasing bradykinin concentration increases NFAT-dependent transcription (Fig. 4A), suggesting that NFAT acts as an integrator capable of producing graded changes in gene expression in response to these convergent signals rather than acting as an all-or-none switch. This idea fits with previous studies that showed NFAT-dependent transcription was regulated in part by the presence of transcriptional partners, such as AP-1 (Groth and Mermelstein, 2003). It is also consistent with the analgesic properties of cyclosporine (Lee et al., 2002a). Furthermore, we noted a 4-fold rightward shift in the concentration-response profile of bradykinin-induced NFAT-dependent transcription (EC50 = 24.2 ± 0.1) relative to the that of bradykinin-evoked [Ca2+]i increases (EC50 = 6.0 ± 0.3). The response profile for bradykinin-evoked [Ca2+]i increases is in close agreement with bradykinin-evoked IP3 production observed previously (Thayer et al., 1988a). Bradykinin B2 receptor activation of PLC increases both IP3 and diacylglycerol. Increased concentrations of bradykinin may be required to activate transcription factors downstream of diacylglycerol production and protein kinase C activation that partner with NFAT to initiate transcription.

    Bradykinin and Prostaglandins in Peripheral Sensitization. The concentration of bradykinin increases during short-(surgery) and long-term (rheumatoid arthritis) inflammation (Hargreaves et al., 1988). Many of the proinflammatory effects of bradykinin in peripheral neurons have been linked to the synthesis and release of prostanoids, including prostaglandin E2 and prostaglandin I2, that are known to sensitize nociceptors (Dray and Perkins, 1993). The rate-limiting step in prostaglandin synthesis is the conversion of arachidonic acid to prostaglandin H2, catalyzed by cyclooxygenase enzymes (Cox-1 and Cox-2). Although Cox-1 is constitutively expressed in many tissues, Cox-2 expression is increased in response to tissue injury and inflammation. The role for cyclooxygenase (and prostanoid) involvement in pain and inflammation is supported by the analgesic properties of nonsteroidal anti-inflammatory drugs. Activation of nociceptors by bradykinin is attenuated by pretreatment with nonsteroidal anti-inflammatory drugs (Dray et al., 1992), further supporting a link between cyclooxygenase activity and bradykinin.

    The mechanisms coupling bradykinin exposure to long-term elevation of prostanoids have received little attention. Cox-2 expression in other tissues is regulated by several Ca2+-activated transcription factors, including cAMP response element-binding protein and nuclear factor-B (Svensson and Yaksh, 2002). Cox-2 expression, driven by calcineurin and NFAT signaling, has been demonstrated in vascular smooth muscle (Robida et al., 2000), spinal neurons (Groth et al., 2007), and T lymphocytes (Iñiguez et al., 2000). Our results demonstrate that bradykinin can elicit B2 receptor and calcineurin-dependent increases in Cox-2 mRNA in sensory neurons. Although these results do not explicitly address whether bradykinin-induced Cox-2 expression is involved in long-term pain syndromes in vivo, they do suggest a potential mechanism whereby short-lived stimuli may contribute to long-term sensitization produced by inflammation.

    Bradykinin activated the transcription factor NFAT in DRG neurons. This activation depended on B2 receptor activation of phospholipase C and subsequent Ca2+ release to trigger calcineurin-dependent NFAT translocation to the nucleus and subsequent NFAT-dependent transcription. Furthermore, our data demonstrate that activation of this cascade leads to increases in mRNA for the proinflammatory enzyme cyclooxygenase-2. This suggests a potential mechanism by which bradykinin may elicit long-lasting changes in the sensitivity of sensory neurons to painful stimuli.

    Acknowledgements

    We thank Dr. Masamitsu Iino for providing the GFP-NFAT4 expression vector and Dr. Paul Mermelstein for providing pNFAT-luciferase.

    ABBREVIATIONS: PLC, phospholipase C; Cox-2, cyclooxygenase-2; BDNF, brain-derived neurotrophic factor; CPA, cyclopiazonic acid; CSA, cyclosporin A; Ct, cycle threshold; DRG, dorsal root ganglion; ER, endoplasmic reticulum; GFP, green fluorescent protein; HHSS, HEPES-Hanks' salt solution; IP3, inositol 1,4,5-triphosphate; NFAT, nuclear factor of activated T-cells; NGF, nerve growth factor; PCR, polymerase chain reaction; TRPV1, transient receptor potential vanilloid receptor 1; RLU, relative light unit; U-73122, 1-[6-[[17β-methoxyestra-1,3,5(10)-trien-17-yl]amino]-hexyl]-1H-pyrrole-2,5-dione; U-73343, 1-[6-[[17β-3-methoxyestra-1,3,5(10)-trien-17-yl]amino]hexyl]-2,5-pyrrolidine-dione; HOE140, icatibant.

    1 Current affiliation: Department of Pharmacology, The University of Iowa, Iowa City, Iowa.

【参考文献】
  Barlas A, Sugio K, and Greenbaum LM (1985) Release of T-kinin and bradykinin in carrageenin induced inflammation in the rat. FEBS Lett 190: 268-270.[CrossRef][Medline]

Beals CR, Sheridan CM, Turck CW, Gardner P, and Crabtree GR (1997) Nuclear export of NF-ATc enhanced by glycogen synthase kinase-3. Science 275: 1930-1934.[Abstract/Free Full Text]

Benedito AB, Lehtinen M, Massol R, Lopes UG, Kirchhausen T, Rao A, and Bonni A (2005) The transcription factor NFAT3 mediates neuronal survival. J Biol Chem 280: 2818-2825.[Abstract/Free Full Text]

Bhave G, Hu HJ, Glauner KS, Zhu W, Wang H, Brasier DJ, Oxford GS, and Gereau RW 4th (2003) Protein kinase C phosphorylation sensitizes but does not activate the capsaicin receptor transient receptor potential vanilloid 1 (TRPV1). Proc Natl Acad Sci U S A 100: 12480-12485.[Abstract/Free Full Text]

Cesare P, Dekker LV, Sardini A, Parker PJ, and McNaughton PA (1999) Specific involvement of PKC-epsilon in sensitization of the neuronal response to painful heat. Neuron 23: 617-624.[CrossRef][Medline]

Chuang HH, Prescott ED, Kong H, Shields S, Jordt SE, Basbaum AI, Chao MV, and Julius D (2001) Bradykinin and nerve growth factor release the capsaicin receptor from PtdIns(4,5)P2-mediated inhibition. Nature 411: 957-962.[CrossRef][Medline]

Crabtree GR and Olson EN (2002) NFAT signaling: choreographing the social lives of cells. Cell 109 (Suppl): S67-S79.[CrossRef][Medline]

Dolmetsch RE, Xu K, and Lewis RS (1998) Calcium oscillations increase the efficiency and specificity of gene expression. Nature 392: 933-936.[CrossRef][Medline]

Dray A, Patel IA, Perkins MN, and Rueff A (1992) Bradykinin-induced activation of nociceptors: receptor and mechanistic studies on the neonatal rat spinal cord-tail preparation in vitro. Br J Pharmacol 107: 1129-1134.[Medline]

Dray A and Perkins M (1993) Bradykinin and inflammatory pain. Trends Neurosci 16: 99-104.[CrossRef][Medline]

Graef IA, Mermelstein PG, Stankunas K, Neilson JR, Deisseroth K, Tsien RW, and Crabtree GR (1999) L-type calcium channels and GSK-3 regulate the activity of NF-ATc4 in hippocampal neurons. Nature 401: 703-708.[CrossRef][Medline]

Graef IA, Wang F, Charron F, Chen L, Neilson J, Tessier-Lavigne M, and Crabtree GR (2003) Neurotrophins and netrins require calcineurin/NFAT signaling to stimulate outgrowth of embryonic axons. Cell 113: 657-670.[CrossRef][Medline]

Groth RD, Coicou LG, Mermelstein PG, and Seybold VS (2007) Neurotrophin activation of NFAT-dependent transcription contributes to the regulation of pronociceptive genes. J Neurochem, in press. doi: 10.1111/j.1471-4159.2007.4632.x.

Groth RD and Mermelstein PG (2003) Brain-derived neurotrophic factor activation of NFAT (nuclear factor of activated T-cells)-dependent transcription: a role for the transcription factor NFATc4 in neurotrophin-mediated gene expression. J Neurosci 23: 8125-8134.[Abstract/Free Full Text]

Grynkiewicz G, Poenie M, and Tsien RY (1985) A new generation of Ca2+ indicators with greatly improved fluorescence properties. J Biol Chem 260: 3440-3450.[Abstract/Free Full Text]

Hargreaves KM, Troullos ES, Dionne RA, Schmidt EA, Schafer SC, and Joris JL (1988) Bradykinin is increased during acute and chronic inflammation: therapeutic implications. Clin Pharmacol Ther 44: 613-621.[Medline]

I?iguez MA, Martinez-Martinez S, Punzon C, Redondo JM, and Fresno M (2000) An essential role of the nuclear factor of activated T cells in the regulation of the expression of the cyclooxygenase-2 gene in human T lymphocytes. J Biol Chem 275: 23627-23635.[Abstract/Free Full Text]

Jackson JG and Thayer SA (2006) Mitochondrial modulation of Ca2+-induced Ca2+-release in rat sensory neurons. J Neurophysiol 96: 1093-1104.[Abstract/Free Full Text]

Kim M, Thayer SA, and Usachev YM (2006) Mitochondria regulates activation of the transcription factor NFAT in DRG neurons. Soc Neurosci Abstr 32: 321.9.

Lee PC, Tsai YC, Hung CJ, Lin YJ, Lei HY, Chuang JI, and Hsu KS (2002a) Induction of antinociception and increased met-enkephalin plasma levels by cyclosporine and morphine in rats: implications of the combined use of cyclosporine and morphine and acute posttransplant neuropsychosis. J Surg Res 106: 1-6.[CrossRef][Medline]

Lee YJ, Zachrisson O, Tonge DA, and McNaughton PA (2002b) Upregulation of bradykinin B2 receptor expression by neurotrophic factors and nerve injury in mouse sensory neurons. Mol Cell Neurosci 19: 186-200.[CrossRef][Medline]

Lo TM and Thayer SA (1993) Refilling the inositol 1, 4, 5-triphosphate-sensitive Ca2+ store in neuroblastoma x glioma hybrid NG108-15 cells. Am J Physiol 33: C641-C653.

Marceau F and Regoli D (2004) Bradykinin receptor ligands: therapeutic perspectives. Nat Rev Drug Discov 3: 845-852.[CrossRef][Medline]

Oh EJ and Weinreich D (2004) Bradykinin decreases K+ and increases Cl- conductances in vagal afferent neurones of the guinea pig. J Physiol 558: 513-526.[Abstract/Free Full Text]

Pfaffl MW (2001) A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res 29: e45.[Abstract/Free Full Text]

Premkumar LS and Ahern GP (2000) Induction of vanilloid receptor channel activity by protein kinase C. Nature 408: 985-990.[CrossRef][Medline]

Robida AM, Xu K, Ellington ML, and Murphy TJ (2000) Cyclosporin A selectively inhibits mitogen-induced cyclooxygenase-2 gene transcription in vascular smooth muscle cells. Mol Pharmacol 58: 701-708.[Abstract/Free Full Text]

Seybold VS, Coicou LG, Groth RD, and Mermelstein PG (2006) Substance P initiates NFAT-dependent gene expression in spinal neurons. J Neurochem 97: 397-407.[CrossRef][Medline]

Shin J, Cho H, Hwang SW, Jung J, Shin CY, Lee SY, Kim SH, Lee MG, Choi YH, Kim J, et al. (2002) Bradykinin-12-lipoxygenase-VR1 signaling pathway for inflammatory hyperalgesia. Proc Natl Acad Sci U S A 99: 10150-10155.[Abstract/Free Full Text]

Shmigol A, Verkhratsky A, and Isenberg G (1995) Calcium-induced calcium release in rat sensory neurons. J Physiol 489: 627-636.[Abstract/Free Full Text]

Snyder SH, Sabatini DM, Lai MM, Steiner JP, Hamilton GS, and Suzdak PD (1998) Neural actions of immunophilin ligands. Trends Pharmacol Sci 19: 21-26.[CrossRef][Medline]

Steranka LR, Manning DC, DeHaas CJ, Ferkany JW, Borosky SA, Connor JR, Vavrek RJ, Stewart JM, and Snyder SH (1988) Bradykinin as a pain mediator: receptors are localized to sensory neurons, and antagonists have analgesic actions. Proc Natl Acad Sci U S A 85: 3245-3249.[Abstract/Free Full Text]

Svensson CI and Yaksh TL (2002) The spinal phospholipase-cyclooxygenase-prostanoid cascade in nociceptive processing. Annu Rev Pharmacol Toxicol 42: 553-583.[CrossRef][Medline]

Thayer SA, Perney TM, and Miller RJ (1988a) Regulation of calcium homeostasis in sensory neurons by bradykinin. J Neurosci 8: 4089-4097.[Abstract]

Thayer SA, Sturek M, and Miller RJ (1988b) Measurement of neuronal Ca2+ transients using simultaneous microfluorimetry and electrophysiology. Pflugers Arch 412: 216-223.[CrossRef][Medline]

Tomida T, Hirose K, Takizawa A, Shibasaki F, and Iino M (2003) NFAT functions as a working memory of Ca2+ signals in decoding Ca2+ oscillation. EMBO J 22: 3825-3832.[CrossRef][Medline]

Usachev YM, DeMarco SJ, Campbell C, Strehler EE, and Thayer SA (2002) Bradykinin and ATP accelerate Ca2+ efflux from rat sensory neurons via protein kinase C and the plasma membrane Ca2+ pump isoform 4. Neuron 33: 113-122.[CrossRef][Medline]

Usachev YM, Khammanivong A, Campbell C, and Thayer SA (2000) Particle-mediated gene transfer to rat neurons in primary culture. Pflugers Arch 439: 730-738.[CrossRef][Medline]

Wang H, Kohno T, Amaya F, Brenner GJ, Ito N, Allchorne A, Ji R-R, and Woolf CJ (2005) Bradykinin produces pain hypersensitivity by potentiating spinal cord glutamatergic synaptic transmission. J Neurosci 25: 7986-7992.[Abstract/Free Full Text]

Woolf CJ and Costigan M (1999) Transcriptional and posttranslational plasticity and the generation of inflammatory pain. Proc Natl Acad Sci U S A 96: 7723-7730.[Abstract/Free Full Text]

Yoshida T and Mishina M (2005) Distinct roles of calcineurin-nuclear factor of activated T-cells and protein kinase A-cAMP response element-binding protein signaling in presynaptic differentiation. J Neurosci 25: 3067-3079.[Abstract/Free Full Text]

作者单位：Department of Pharmacology, University of Minnesota Medical School, Minneapolis, Minnesota

【关键词】  Brain-Derived

    Brain-derived neurotrophic factor (BDNF), a member of the neurotrophin family, plays an important role in synaptic plasticity. In this issue of Molecular Pharmacology, Ou and Gean (p. 350) thoroughly describe the molecular cascade by which fear learning leads to an increase in BDNF expression in the lateral amygdala (LA). Calcium influx through N-methyl-D-aspartate receptors and L-type voltage-dependent calcium channels, which occurs in the LA during fear conditioning, activates protein kinase A and Ca2+/calmodulin-dependent protein kinase IV. Each induces phosphorylation of cAMP response element-binding protein, which binds to the BDNF promoter, leading to BDNF expression in the LA, and contributes to fear memory consolidation.

    The activity-dependent modification of synapses, a process known as synaptic plasticity, permits the brain to generate efficient neural networks that facilitate advantageous behavioral adaptations. Although an extensive body of research has demonstrated the importance of synaptic strengthening in the beneficial functions of learning and memory, recent evidence also suggests that a host of pathologic conditions, including mood disorders and drug addiction, engage overlapping mechanisms. Thus, the elucidation of molecular pathways involved in driving these physiological changes has broad clinical implications (Malenka and Bear, 2004; MacKinnon and Zamoiski, 2006).

    The neurotrophins, a family of structurally related proteins known for their role in promoting neuronal differentiation and survival during development (Levi-Montalcini, 1987; Leibrock et al., 1989; Barde, 1994), have recently surfaced as playing an important role in mediating synaptic plasticity (Schinder and Poo, 2000; Lu, 2003, 2004; Bramham and Messaoudi, 2005; Arancio and Chao, 2007). In this issue of Molecular Pharmacology, Ou and Gean (2007) report on their investigation of the mechanism by which one member of this family, brain-derived neurotrophic factor (BDNF), mediates fear memory consolidation in the amygdala. BDNF and its main receptor, TrkB, are fast emerging as major regulators of synaptic transmission and plasticity in the adult brain. In mammals, BDNF is synthesized, stored, and released from excitatory neurons containing glutamate (Lessmann et al., 2003) and, in some populations, dopamine (Berton et al., 2006).

    Past investigations into the role of BDNF in plasticity have predominantly focused on the hippocampus, a structure traditionally associated with learning and memory. In that structure, TrkB receptors have been specifically localized to the pre- and postsynaptic elements of glutamatergic synapses (Drake et al., 1999) and coimmunoprecipitate with the transmembrane NMDA receptor protein, which allows calcium entry into the cell in response to detection of coincidental rises in both presynaptic glutamate and postsynaptic voltage (Aoki et al., 2000). Because this calcium influx is necessary for the initiation of cellular changes, the colocalization of NMDA receptors with BDNF and its receptor, TrkB, at synaptic junctions sets the stage to synchronize bidirectional synaptic optimization.

    Much evidence indicates that learning initiates alterations in glutamate-dependent excitatory synaptic transmission that subsequently stabilize through structural changes at postsynaptic sites on dendritic spines (for review, see Lamprecht and LeDoux, 2004). Direct infusion of BDNF into the hippocampus enhances synaptic strength, both in vitro (Kang and Schuman, 1995; Levine et al., 1995) and in vivo (Messaoudi et al., 1998), as well as modulates the induction of long-term potentiation (Patterson et al., 1996; Messaoudi et al., 2002) and structural changes in dendritic spines (Alonso et al., 2004). Recent work by Tyler et al. (2002) also suggests that BDNF activation is necessary for the learning-induced modification of hippocampal spines, which may also involve the activation of TrkB (von Bohlen und Halbach, 2006). Furthermore, Bekinschtein et al. (2007) recently demonstrated that BDNF-dependent storage of long-term memories occurs within hours after acquisition of an associative learning task, suggesting that BDNF is likely to be involved in memory stabilization.

    Despite the traditional popularity of the hippocampus in all matters of learning and memory, there is increasing empirical support for the role of another structure—the amygdala—in the types of synaptic changes facilitated by BDNF. In particular, mounting evidence now indicates a role for BDNF signaling in the basal and lateral nuclei of the amygdala (Rattiner et al., 2004a; Ou and Gean, 2006), areas known to be necessary for the formation of learned fear associations (LeDoux, 2000). Because the amygdala has been implicated in many pathologic conditions, including post-traumatic stress (Garakani et al., 2006), anxiety (Rauch et al., 2006), and autism spectrum disorders (Baron-Cohen et al., 2000; Bachevalier and Loveland, 2006), considerable efforts have been devoted to the characterization of this circuitry as a central site for emotion-induced neuronal plasticity (LeDoux, 2000; Maren, 2001; Paré et al., 2004; Wilensky et al., 2006). A number of studies have shown, using Pavlovian fear conditioning paradigms, that the physiological basis for such changes begins with the relay of sensory information from the medial geniculate nucleus of the thalamus to the lateral amygdala (LA), where the initial association is made via an LTP-like mechanism, followed by the intra-amygdala transfer of signals to the central nucleus of the amygdala, which facilitates the expression of a fear response by way of projections to brainstem and hypothalamic targets (Davis, 1997; LeDoux, 2000; Paré et al., 2004). Together, these findings support the notion that changes in synaptic strength are required for the acquisition of emotional memories. At a more profound level, however, our knowledge of these larger-scale anatomical modifications remains bound by our more limited comprehension of the molecular machinery governing those changes.

    A recent study found temporally specific increases in BDNF gene expression to occur in the basal/lateral portion of the amygdala (BLA) after paired stimuli that supported learning, but not after exposure to neutral or aversive stimuli alone (Rattiner et al., 2004a). BDNF signaling through TrkB receptors was also found to be necessary for the consolidation of fear memories (Rattiner et al., 2004a). In agreement with the findings of Rattiner et al. (2004a), Ou and Gean (2006) reported increases in BDNF protein expression and activation of TrkB receptors in the amygdala. Their study further revealed that intra-amygdala infusion of a TrkB ligand scavenger or the inhibition of Trk receptors impaired fear memory assessed 24 h after training. In addition, they showed that BDNF phosphorylates mitogen-activated protein kinase (MAPK), and this is blocked by the Trk receptor inhibitor K252a (Ou and Gean, 2006). The BDNF-induced phosphorylation of MAPK occurs via Shc binding to the TrkB receptor, which leads to the activation of Ras, Raf, MEK, and MAPK. BDNF also phosphorylates MAPK via activation of PI-3 kinase (Ou and Gean, 2006).

    In this issue of Molecular Pharmacology, Ou and Gean (2007) extend their earlier work to a characterization of the molecular cascades underlying fear conditioning that exert transcriptional and translational control over BDNF expression in the amygdala. Consistent with the previous work of Rattiner et al. (2004b), they show a significant increase in BDNF exon I- and III-containing mRNA in the amygdala of fear-conditioned rats. Inhibition of protein synthesis and translation, using intra-LA anisomycin or actinomycin D, respectively, attenuates this increase in fear-conditioning-induced BDNF expression. Furthermore, they demonstrate that the increase in BDNF depends on the activation of NMDA receptors as well as L-type voltage-dependent calcium channels (L-VDCC), the blockade of which significantly attenuates BDNF expression. A similar reduction was also apparent after the pharmacological inhibition of PKA and CaMKIV activity. In addition, through the use of DNA affinity precipitation and chromatin immunoprecipitation assays, Ou and Gean (2007) demonstrate a specific increase in the binding of phosphorylated cAMP response element binding protein (CREB) to exon I and III promoters after fear conditioning. They found that sequestration of endogenous BDNF during fear conditioning by infusion of a TrkB IgG did not affect the BDNF protein level increases typically observed 1 h after conditioning. This suggests that whereas BDNF signaling through TrkB receptors in the amygdala is required for long-term memory (Rattiner et al., 2004a; Ou and Gean, 2006), it is not necessary to regulate the increase in BDNF protein levels induced by fear conditioning (Ou and Gean, 2007).

    Many of the most common psychiatric disorders that afflict humans are emotional disorders, a number of which involve the activation of fear circuitry in the brain. To develop suitable treatments for anxiety-related disorders, it is necessary to develop a better understanding of the molecular mechanisms that underlie their development and manifestation. Ou and Gean's (2007) findings elegantly illustrate that calcium influx through NMDA receptors and L-VDCC channels, known to occur during fear conditioning, activates PKA and CaMKIV, each inducing CREB phosphorylation. In turn, phosphorylated CREB binds to the BDNF promoter, leading to an increase in BDNF expression in the amygdala, and probably contributes to fear memory consolidation (see Fig. 1). Ou and Gean (2007) describe a tight molecular cascade linking the initial physiological events that take place during fear conditioning to the expression of BDNF—a potent modulator of synaptic plasticity that could lead to the restructuring of synapses in the LA. Taken together, these findings implicate the BDNF signaling cascade in the amygdala as a potential target for novel pharmacological interventions.

    Fig. 1. BDNF signaling cascade involved during fear conditioning. Calcium influx through NMDA receptors and L-VDCC channels, which occurs in the LA during fear conditioning, activates adenyl cyclase (AC) and PKA. Activated PKA translocates to the nucleus and induces CREB phosphorylation. Increase in intracellular calcium also activates CaMKIV and leads to phosphorylation of CREB at Ser-133. Activated CREB binds to the BDNF promoter, leading to BDNF expression in the LA, and contributes to fear memory consolidation (from Ou and Gean, 2007). Fear conditioning is also associated with binding of BDNF to TrkB receptors. This results in the association of Shc and TrkB receptor [structure of ligand binding domain after the work of Ultsch et al. (1999)], and leads to the activation of Ras, Raf, MEK, and MAPK (not shown). BDNF also phosphorylates MAPK via activation of PI3 kinase (Ou and Gean, 2006) (not shown).

    ABBREVIATIONS: BDNF, brain-derived neurotrophic factor; NMDA, N-methyl-D-aspartate; LA, lateral amygdala; LTP, long-term potentiation; MAPK, mitogen-activated protein kinase; MEK, mitogen-activated protein kinase kinase; L-VDCC, L-type voltage-dependent calcium channel; PKA, protein kinase A; CaMKIV, Ca2+/calmodulin-dependent protein kinase IV; CREB, cAMP response element-binding protein.

【参考文献】
  Aoki C, Wu K, Elste A, Len G, Lin S, McAuliffe G, and Black IB (2000) Localization of brain-derived neurotrophic factor and TrkB receptors to postsynaptic densities of adult rat cerebral cortex. J Neurosci Res 59: 454-463.[CrossRef][Medline]

Alonso M, Medina JH, Pozzo-Miller L (2004) ERK1/2 activation is necessary for BDNF to increase dendritic spine density in hippocampal CA1 pyramidal neurons. Learn Mem 11: 172-178.[Abstract/Free Full Text]

Arancio O and Chao MV (2007) Neurotrophins, synaptic plasticity and dementia. Curr Opin Neurobiol, in press.

Bachevalier J and Loveland KA (2006) The orbitofrontal-amygdala circuit and self-regulation of social-emotional behavior in autism. Neurosci Biobehav Rev 30: 97-117.[CrossRef][Medline]

Baron-Cohen S, Ring HA, Bullmore ET, Wheelwright S, Ashwin C, and Williams SC (2000) The amygdala theory of autism. Neurosci Biobehav Rev 24: 355-364.[CrossRef][Medline]

Barde (1994) Neurotrophins: a family of proteins supporting the survival of neurons. Prog Clin Biol Res 390: 45-56.[Medline]

Bekinschtein P, Cammarota M, Igaz LM, Bevilaqua LR, Izquierdo I, and Medina JH (2007) Persistence of long-term memory storage requires a late protein synthesis- and BDNF-dependent phase in the hippocampus. Neuron 53: 261-277.[CrossRef][Medline]

Berton O, McClung CA, Dileone RJ, Krishnan V, Renthal W, Russo SJ, Graham D, Tsankova NM, Bolanos CA, Rios M, et al. (2006) Essential role of BDNF in the mesolimbic dopamine pathway in social defeat stress. Science 311: 864-868.[Abstract/Free Full Text]

Bramham CR and Messaoudi E (2005) BDNF function in adult synaptic plasticity: the synaptic consolidation hypothesis. Prog Neurobiol 76: 99-125.[CrossRef][Medline]

Davis M (1997) Neurobiology of fear responses: the role of the amygdala. J Neuropsychiatry Clin Neurosci 9: 382-402.[Abstract/Free Full Text]

Drake CT, Milner TA, and Patterson SL (1999) Ultrastructural localization of full-length trkB immunoreactivity in rat hippocampus suggests multiple roles in modulating activity-dependent synaptic plasticity. J Neurosci 19: 8009-8026.[Abstract/Free Full Text]

Garakani A, Mathew SJ, and Charney DS (2006) Neurobiology of anxiety disorders and implications for treatment. Mt Sinai J Med 73: 941-949.[Medline]

Kang H and Schuman EM (1995) Long-lasting neurotrophin-induced enhancement of synaptic transmission in the adult hippocampus. Science 267: 1658-1662.[Abstract/Free Full Text]

Lamprecht and LeDoux (2004) Structural plasticity and memory. Nat Rev Neurosci 5: 45-54.[CrossRef][Medline]

LeDoux JE (2000) Emotion circuits in the brain. Annu Rev Neurosci 23: 155-184.[CrossRef][Medline]

Leibrock J, Lottspeich F, Hohn A, Hofer M, Hengerer B, Masiakowski P, Thoenen H, and Barde YA (1989) Molecular cloning and expression of brain-derived neurotrophic factor. Nature 341: 149-152.[CrossRef][Medline]

Lessmann V, Gottmann K, and Malcangio M (2003) Neurotrophin secretion: current facts and future prospects. Prog Neurobiol 69: 341-374.[CrossRef][Medline]

Levi-Montalcini R (1987) The nerve growth factor 35 years later. Science 237: 1154-1162.[Free Full Text]

Levine ES, Dreyfus CF, Black IB, and Plummer MR (1995) Brain-derived neurotrophic factor rapidly enhances synaptic transmission in hippocampal neurons via postsynaptic tyrosine kinase receptors. Proc Natl Acad Sci U S A 92: 8074-8077.[Abstract/Free Full Text]

Lu B (2003) BDNF and activity-dependent synaptic modulation. Learn Mem 10: 86-98.[Abstract/Free Full Text]

Lu B (2004) Acute and long-term synaptic modulation by neurotrophins. Prog Brain Res 146: 137-150.[Medline]

MacKinnon and Zamoiski (2006) Panic comorbidity with bipolar disorder: what is the manic-panic connection? Bipolar Disord 8: 648-664.[CrossRef][Medline]

Malenka RC and Bear MF (2004) LTP and LTD: an embarrassment of riches. Neuron 44: 5-21.[CrossRef][Medline]

Maren S (2001) Neurobiology of Pavlovian fear conditioning. Annu Rev Neurosci 24: 897-931.[CrossRef][Medline]

Messaoudi E, Bardsen K, Srebro B, and Bramham CR (1998) Acute intrahippocampal infusion of BDNF induces lasting potentiation of synaptic transmission in the rat dentate gyrus. J Neurophysiol 79: 496-499.[Abstract/Free Full Text]

Messaoudi E, Ying SW, Kanhema T, Croll SD, and Bramham CR (2002) Brain-derived neurotrophic factor triggers transcription-dependent, late phase long-term potentiation in vivo. J Neurosci 22: 7453-7461.[Abstract/Free Full Text]

Ou LC and Gean PW (2006) Regulation of amygdala-dependent learning by brain-derived neurotrophic factor is mediated by extracellular signal-regulated kinase and phosphatidylinositol-3-kinase. Neuropsychopharmacology 31: 287-296.[CrossRef][Medline]

Ou LC and Gean PW (2007) Transcriptional regulation of brain-derived neurotrophic factor in the amygdala during consolidation of fear memory. Mol Pharmacol 72: 350-358.[Abstract/Free Full Text]

Paré D, Quirk GJ, and LeDoux JE (2004) New vistas on amygdala networks in conditioned fear. J Neurophysiol 92: 1-9.[Abstract/Free Full Text]

Patterson SL, Abel T, Deuel TA, Martin KC, Rose JC, and Kandel ER (1996) Recombinant BDNF rescues deficits in basal synaptic transmission and hippocampal LTP in BDNF knockout mice. Neuron 16: 1137-1145.[CrossRef][Medline]

Rattiner LM, Davis M, French CT, and Ressler KJ (2004a) Brain-derived neurotrophic factor and tyrosine kinase receptor B involvement in amygdala-dependent fear conditioning. J Neurosci 24: 4796-4806.[Abstract/Free Full Text]

Rattiner LM, Davis M, and Ressler KJ (2004b) Differential regulation of brain-derived neurotrophic factor transcripts during the consolidation of fear learning. Learn Mem 11: 727-731.[Abstract/Free Full Text]

Rauch SL, Shin LM, and Phelps EA (2006) Neurocircuitry models of posttraumatic stress disorder and extinction: human neuroimaging research-past, present, and future. Biol Psychiatry 60: 376-382.[CrossRef][Medline]

Schinder AF and Poo M (2000) The neurotrophin hypothesis for synaptic plasticity. Trends Neurosci 23: 639-645.[CrossRef][Medline]

Tyler WJ, Alonso M, Bramham CR, and Pozzo-Miller LD (2002) From acquisition to consolidation: on the role of brain-derived neurotrophic factor signaling in hippocampal-dependent learning. Learn Mem 9: 224-237.[Abstract/Free Full Text]

Ultsch MH, Wiesmann C, Simmons LC, Henrich J, Yang M, Reilly D, Bass SH, and de Vos AM (1999) Crystal structures of the neurotrophin-binding domain of TrkA, TrkB and TrkC. J Mol Biol 290: 149-159.[CrossRef][Medline]

von Bohlen und Halbach O, Krause S, Medina D, Sciarretta C, Minichiello L, and Unsicker K (2006) Regional- and age-dependent reduction in trkB receptor expression in the hippocampus is associated with altered spine morphologies. Biol Psychiatry 59: 793-800.[CrossRef][Medline]

Wilensky AE, Schafe GE, Kristensen MP, and LeDoux (2006) Rethinking the fear circuit: the central nucleus of the amygdala is required for the acquisition, consolidation, and expression of Pavlovian fear conditioning. J Neurosci 26: 12387-12396.[Abstract/Free Full Text]

作者单位：Center for Neural Science, New York University, New York, New York

【关键词】  Quercetin

    We investigated a molecular mechanism underlying quercetin-mediated amelioration of colonic mucosal injury and analyzed chemical structure contributing to the quercetin's effect. Quercetin up-regulated vascular endothelial growth factor (VEGF), an ulcer healing factor, not only in colon epithelial cell lines but also in the inflamed colonic tissue. VEGF derived from quercetin-treated colon epithelial cells promoted tube formation. The VEGF induction was dependent on quercetin-mediated hypoxia-inducible factor-1 (HIF-1) activation. Quercetin delayed HIF-1 protein disappearance, which occurred by inhibiting HIF-prolyl hydroxylase (HPH), the key enzyme for HIF-1 hydroxylation and subsequent von Hippel Lindau-dependent HIF-1 degradation. HPH inhibition by quercetin was neutralized significantly by an elevated dose of iron. Consistent with this, cellular induction of HIF-1 by quercetin was abolished by pretreatment with iron. Two iron-chelating moieties in quercetin, -OH at position 3 of the C ring and/or -OH at positions 3' and 4' of the B ring, enabled the flavonoid to inhibit HPH and subsequently induce HIF-1. Our data suggest that the clinical effect of quercetin may be partly attributed to the activation of an angiogenic pathway HIF-1-VEGF via inhibiting HPH and the chelating moieties of quercetin were required for inhibiting HPH.

    The pathogenesis of both ulcerative colitis and Crohn's disease is unknown, but these forms of inflammatory bowel disease (IBD) may be associated with an inability of the intestinal mucosa to protect itself from luminal challenges and/or inappropriate repair after intestinal injury (Beck and Podolsky, 1999). In general, the gastrointestinal mucosa has a remarkable ability to repair damage. When the integrity of the superficial mucosa is breached, ulcer healing, a complex and tightly regulated process of filling the mucosal defect, begins (Mammen and Matthews, 2003). Angiogenesis, the formation of new microvessels, is one of the critical components for repair of mucosal defect. This facilitates nutrient and oxygen delivery to the injured area, thus enabling cell proliferation and migration (Tarnawski, 2005). A recent study demonstrates that stimulation of vascular factors with vascular endothelial growth factor (VEGF), a potent angiogenic factor, is sufficient for healing of gastrointestinal ulcers, including IBD (Tarnawski, 2006).

    Hypoxia-inducible factor (HIF-1) is a heterodimeric transcription factor composed of HIF-1 and aryl hydrocarbon receptor nuclear translocator (ARNT, HIF-1) (Wang et al., 1995). HIF-1 and HIF-1 mRNAs are constantly expressed under normoxic and hypoxic conditions (Wiener et al., 1996). However, HIF-1 protein is significantly increased by hypoxia, whereas the HIF-1 protein remains constant regardless of oxygen tension (Salceda and Caro, 1997). Under normoxia, HIF-1 protein is remarkably unstable, and its degradation by the proteasome is orchestrated by the ubiquitin protein ligase VHL (Salceda and Caro, 1997; Huang et al., 1998; Cockman et al., 2000; Tanimoto et al., 2000). Under normoxia, VHL recognizes HIF-1 as a substrate because of the enzymatic modification of HIF-1 by prolyl hydroxylases, whose function is inhibited during hypoxia (Ivan et al., 2001; Jaakkola et al., 2001). Stabilization of HIF-1 by hypoxia or its mimetics is accompanied by its nuclear translocation, heterodimerization with HIF-1, and transcription of genes encoding proteins functioning to increase angiogenesis and promote cell survival and proliferation that are physiological responses not only for adaptation to hypoxia but also for repair of damaged tissue (Semenza, 1998). In fact, HIF-1 is identified as a critical factor for barrier protection during mucosal insult and, further, plays a key role in healing gastrointestinal ulcer via transactivating VEGF (Baatar et al., 2002; Hashimoto et al., 2004).

    Quercetin, the aglycone of rutin, is the most common flavonoid in nature. It has numerous biological activities, including anti-inflammatory effect (Nijveldt et al., 2001). In a previous report, we showed that quercetin is effective in ameliorating the experimental colitis of rats and suggested that a molecular mechanism underlying the clinical effect of quercetin is suppression of a major proinflammatory pathway, tumor necrosis factor-dependent NFB activation (Kim et al., 2005). In this study, we investigated an additional molecular mechanism for the clinical effect of quercetin. Our data demonstrate that quercetin up-regulated an ulcer-healing factor, VEGF, in colon epithelial cells and the inflamed colonic tissue, and quercetin-mediated VEGF induction is dependent on HIF-1 activation, which occurs by quercetin inhibition of HIF-prolyl hydroxylase. Furthermore, it was revealed that two chelating moieties in quercetin were required for inhibiting the enzyme and consequently activating HIF-1. These results imply that, in addition to suppressing an inflammation-provoking factor, NFB pathway, quercetin activation of an angiogenic pathway, HIF-1-VEGF, contributes to the clinical effect of quercetin on experimental colitis.

    Chemicals and Animals. 3,5,7,3',4'-Pentahydroxyflavone (quercetin), 3-hydroxyflavone, and 2,4,6-trinotrobenzene-sulfonic acid (TNBS) were purchased from Tokyo Kasei Kogyo Co. (Tokyo, Japan). 5-Hydroxyflavone and 3',4'-dihydroxyflavone, methylated quercetin derivatives, were purchased from Indofine Chemical Co. Inc. (Hillsborough, NJ). 2-Ketoglutarate, ascorbate, (+)-5,6-O-isopropylidene-L-ascorbic acid, and ferrous chloride were purchased from Sigma Chemical Co. (St. Louis, MO). All other chemicals were reagent-grade, commercially available products. Male Sprague-Dawley rats (240260 g, 8 weeks old) were purchased from Daehan Biotec Co. Ltd. (Daegu, Korea) and housed in the animal care facility at Pusan National University (Busan, Korea). The animal experiments were carried out in accordance with the Guide for the Care and Use of Laboratory Animals as adopted and promulgated by the U.S. National Institutes of Health.

    Cell Culture and Transient Transfection. Human colon epithelial cell lines HCT116 and SW620 (American Type Culture Collection, Manassas, VA) were grown in Dulbecco's modified Eagle's medium (Biofluids, Rockville, MD) supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA) and penicillin/streptomycin (Biofluids). For transient transfection of plasmids, 293 cells were plated in 6-cm dishes to be 50 to 60% confluent on the day of cotransfection with Flag-VHL (5 µg; a gift from Dr. J. Issacs, Medical University of South Carolina, Charleston, SC), HA-HIF-1 plasmid (5 µg; a gift from Dr. L. Neckers, National Cancer Institute, Bethesda, MD), and CMV Renilla reniformis luciferase plasmid (4 ng; Promega, Madison, WI). Fugene (Roche, South San Francisco, CA) was used as a transfection reagent. One day after transfection, cells were treated with each reagent as indicated in the figure legends.

    Immunoblot Analysis and Immunoprecipitation. Cells were lysed, and nuclear extracts were prepared as described previously (Andrews and Faller, 1991). To prepare tissue nuclear extracts, the inflamed distal colon was removed and mixed with 5-fold amount of buffer C (10 mM HEPES, pH 7.9, 10 mM KCl, 0.2 mM EDTA, 0.3 µM aprotinin, 1 µM pepstatin, and 1 mM PMSF) followed by homogenation. Nonidet P-40 (10%) was added to the homogenates at the ratio of 50 µl/ml after 20-min incubation in ice, and the mixture was vortexed vigorously for 15 s and centrifuged at 14,000 rpm and 4°C for 3 min to afford the nuclear pellets. After removing the supernatants, an appropriate volume of buffer containing 20 mM HEPES, pH 7.9, 0.4 M NaCl, 1 mM EDTA, 0.3 µM aprotinin, 1 µM pepstatin, and 1 mM PMSF was added to the nuclear pellets, and the tubes were rotated on a small rotatory shaker at 4°C for 20 min followed by centrifugation at 14,000 rpm and 4°C for 10 min. Protein concentration in the supernatants was determined by the BCA method. The nuclear extracts were transferred to a fresh tube and stored at –70°C until used. Cell or tissue nuclear extracts were electrophoretically separated using 7.5 or 10%. Proteins were transferred to nitrocellulose membranes (Protran; Whatman Schleicher and Schuell, Keene, NH), and HIF-1 protein was detected in nuclear extracts using a monoclonal anti-HIF-1 antibody (Transduction for human HIF-1 or Novus for murine HIF-1). Peroxidase-conjugated anti-rabbit or anti-mouse secondary antibody (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK) was used at a dilution of 1:1000. Signals were visualized using the SuperSignal chemiluminescence substrate (Pierce, Rockford, IL). Experiments were performed in duplicate and normalized with antibodies to topoisomerase II (Santa Cruz Biotechnology, Santa Cruz, CA). For immunoprecipitation, 293 cell lysates (0.7 mg of protein), prepared as described in the previous section, were incubated with 20 µl of anti-HA antibody bound beads (Covance Research Products, Berkeley, CA). The beads were washed five times with lysis buffer, resuspended in 1x SDS sample buffer, and boiled for 5 min. Immunoprecipitated proteins were separated by 10% SDS-PAGE. Immunoblot analysis was done as aforementioned.

    In Vitro VHL Capture Assay. Biotinylated wild-type or proline-hydroxylated peptides (corresponding to HIF residues 556–574) were synthesized (American Peptide Company, Sunnyvale, CA), dissolved in sterile water (500 µg/ml), and incubated with streptavidin beads (Pierce ImmunoPure) at 4°C for 2 h. The beads were washed twice with VHL binding buffer (20 mM Tris, pH 8, 100 mM NaCl, 1 mM EDTA, and 0.5% Nonidet P-40) and three times with reaction buffer (20 mM Tris, pH 7.5, 5 mM KCl, 1.5 mM MgCl2, and 100 µM dithiothreitol). For each condition, 2 µg of peptide/20 µl of beads was aliquoted into separate tubes, and reaction buffer was added, along with cofactors (100 µM 2-ketoglutaric acid, 100 µM L-ascorbic acid, and 50 µM ferrous chloride). The beads and HPH cofactors were mixed at room temperature for 15 min in reaction buffer. Before this incubation, any inhibitors or competing factors were added to the appropriate tubes. Separate in vitro-translated (IVT) reactions (Promega) were the source for the HIF prolyl hydroxylase protein (HPH-2 plasmid was kindly provided by S. McKnight, University of Texas Medical Center, Dallas, TX) and Flag-VHL (Isaacs et al., 2002). A 5-µl aliquot of IVT HPH-2 was added to the bead-peptide mixture for 1 h at 30°C. Afterward, the beads were washed with VHL binding buffer and 10 µl of Flag-VHL IVT was added to the beads overnight at 4°C. The beads were washed, SDS Laemmli buffer was added, the samples were boiled, subjected to SDS-PAGE, and resultant blots were probed for Flag.

    VEGF Analysis. Cells were treated as indicated in the figure legends. Medium was collected after 10-h treatment. A VEGF ELISA kit (R&D Systems, Minneapolis, MN) was used to assess secreted VEGF levels from a 200-µl aliquot of medium. Each sample was harvested for protein, which was used to normalize VEGF levels. An experiment for each condition was carried out in triplicate. To measure VEGF level in the tissue, the inflamed distal colon was removed and mixed with 5-fold amount of buffer C (10 mM HEPES, 10 mM KCl, 0.1 mM EDTA, 0.3 µM aprotinin, 1 µM pepstatin, and 1 mM PMSF) followed by homogenization. The homogenates were centrifuged at 2000 rpm and 4°C for 2 min. The supernatant (100 µl) was transferred to a fresh microtube and then was centrifuged again at 14,000 rpm and 4°C for 10 min. An appropriate volume of the supernatant was subjected to VEGF ELISA.

    Tube Formation Assay. HUVEC cells (5 x 105 cells) were seeded on a layer of previously polymerized Matrigel. The culture supernatants (1/10 volume of HUVEC media) obtained from HCT cells, left untreated or treated with quercetin (25 µM), were added to each well and incubated at 37°C. VEGF was used as a positive control. After 24 h, changes of cell morphology were captured through a phase-contrast microscope and photographed. The same experiment was done in the presence of a VEGF-neutralizing antibody (1 µg/ml, R&D Systems).

    TNBS-Induced Inflammation. Inflammation was induced by the method of Morris et al. (1989) and Yano et al. (2002). In brief, before induction of colitis, rats were starved for 24 h but had free access to water. The rats were lightly anesthetized with ether. A rubber cannula (outer diameter, 2 mm) was inserted rectally into the colon such that the tip was 8 cm proximal to the anus, approximately at the splenic flexure. TNBS dissolved in 50% (v/v) aqueous ethanol was instilled into the colon via the rubber cannula (15 mg/0.3 ml/rat).

    Fig. 1. Quercetin increased secretion of an angiogenic factor, VEGF. A, colon epithelial cells (HCT116 and SW620) were treated with various concentrations of quercetin for 10 h, and VEGF in the cell culture supernatants was analyzed as described under Materials and Methods. *, P < 0.05, and **, P < 0.01 versus control. B, various concentrations of quercetin were administered rectally to inflamed rats once a day for 6 days. Four hours after the last (sixth) treatment with quercetin, the inflamed distal colon was removed, diluted with buffer C, homogenized, and centrifuged. VEGF in the supernatants was analyzed as described under Materials and Methods. *, P < 0.05, and **, P < 0.01 versus control. The data in A and B are means ± S.E. (n = 3).

    Data Analysis. Results were expressed as means ± S.E. The statistical differences among the results of the various groups were compared by the Student's t test. A value of p < 0.05 was considered significant.

    Quercetin Up-Regulates VEGF in Human Colon Epithelial Cells and the Inflamed Colonic Tissue of Rats. We reported previously that quercetin is effective in ameliorating TNBS-induced rat colitis, and the clinical effect is elicited by at least partly inhibiting a major pro-inflammatory NFB pathway (Kim et al., 2005). To explore an additional pharmacological mechanism of the quercetin effect on experimental colitis, we investigated whether quercetin, in addition to suppressing an inflammatory signal, could stimulate a tissue repair signal, which accelerates healing of inflammatory injury. Because quercetin is able to induce VEGF in endothelial cells (Wilson and Poellinger, 2002), and VEGF promotes angiogenesis, a critical step for tissue repair, we examined whether quercetin increased secretion of VEGF in human colon epithelial cells. Cells were treated with quercetin for 10 h. The cell culture supernatants were subjected to VEGF ELISA. As shown in Fig. 1A, quercetin induced VEGF in the cells in a dose-dependent manner. To see whether quercetin is also able to induce VEGF in the inflamed colonic tissue, a 300-µl aliquot of quercetin (50 and 100 µM) in pH 6.8 phosphate-buffered saline buffer was administered to the inflamed site through rectal route once a day starting from 1 day after induction of inflammation by TNBS. The inflamed colonic tissues were removed and homogenized 4 h after the last (sixth) treatment with quercetin followed by centrifugation. VEGF levels in the supernatants were measured using a VEGF ELISA kit. As shown in Fig. 1B, quercetin increased VEGF levels in the inflamed colonic tissue up to approximately 2-fold.

    VEGF Induction by Quercetin Is Dependent on HIF-1. Because VEGF is a target gene of HIF-1 (Semenza, 1998), we examined whether VEGF induction by quercetin was mediated by HIF-1. First, we tested whether quercetin was able to up-regulate HIF-1 protein in cells and the inflamed tissue. Colon epithelial cells were treated with quercetin and were lysed to obtain nuclear extracts. HIF-1 levels were examined by Western blot. As shown in Fig. 2A, quercetin up-regulated HIF-1 in the colon epithelial cells in a dose-dependent manner. To examine HIF-1 induction by quercetin in the inflamed colonic tissue, quercetin was administered rectally after induction of inflammation as aforementioned. The inflamed distal colon was subjected to nuclear extraction, and HIF-1 levels in the nuclear extracts were monitored by Western blot. As shown in Fig. 2B, HIF-1 protein was up-regulated in the inflamed colonic tissue. These results suggest that HIF-1 up-regulation is correlated with VEGF induction upon quercetin treatment. To further clarify that quercetin induction of VEGF is dependent on HIF-1, quercetin was treated in hepa1c1c7 cells that contain wild-type ARNT and in matched hepa1c4 cells that are unable to transactivate HIF-1-dependent genes because of a genetic defect in ARNT (Li et al., 1996), and ELISA and Western blotting were performed to detect VEGF in the cell culture supernatants and HIF-1 protein in the nuclear extracts. Hypoxia (approximately 1% O2) was used as a positive control for VEGF induction. As shown Fig. 2C, hypoxia and quercetin induced VEGF only in hepa1c1c7 cells. Furthermore, this increase in VEGF secretion was correlated with an increase in HIF-1 level. However, in ARNT-deficient hepa1c4 cells, quercetin failed to increase VEGF secretion, thereby demonstrating that quercetin-mediated VEGF induction is dependent on transcriptionally active HIF-1. Because quercetin activated HIF-1-VEGF, an angiogenic pathway, we wondered whether quercetin could promote angiogenesis via the pathway. To examine this, HCT116 cells were treated with quercetin for 10 h, and the cell culture supernatants were added to the wells in which endothelial cells were seeded, and the tube formation was measured. VEGF was used as a positive control. To see whether quercetin-induced VEGF is the main factor for the in vitro tube formation, the same experiment was carried out in the presence of a VEGF-neutralizing antibody. As shown in Fig. 2D, as predicted, VEGF treatment promoted tube formation, and the VEGF antibody neutralized the effect of VEGF. The cell culture supernatant obtained from HCT 116 cells left untreated with quercetin showed tube formation similar to that of a control (no addition of the cell culture supernatant), and the VEGF antibody did not significantly affect tube formation in this condition. On the contrary, the cell culture supernatant obtained from quercetin-treated HCT116 cells enhanced tube formation, which was inhibited significantly by pretreatment with the VEGF antibody. Quantitative analysis of tube formation is shown in Supplemental Data 1A. In vitro cell proliferation assay was also done with the cell culture supernatants. As shown in Supplemental Data 1B, proliferation of HUVEC cells was promoted by the supernatant obtained from quercetin-treated cells, which was comparable with that by VEGF.

    Fig. 2. Quercetin induction of VEGF is dependent on HIF-1. A, colon epithelial cells (HCT116 and SW620) were treated with various concentrations of quercetin for 4 h, and HIF-1 levels were monitored in the nuclear extracts as described under Materials and Methods. B, the tissue pellets that were obtained from Fig. 1B were subjected to nuclear extraction. HIF-1 levels were monitored in the nuclear extracts as described under Materials and Methods. C, hepa1c1c7 cells (H1C1C7) were treated with quercetin (50 µM) or hypoxia for 10 h, and VEGF in the cell culture supernatants and HIF-1 in the nuclear extracts were analyzed as described under Materials and Methods. The same experiment was done using hepa1c4 cells (H1C4) with genetically defective ARNT (except for immunodetecting HIF-1). The data are means ± S.E. (n = 3). *, P < 0.01 versus control. D, HUVEC cells (5 x 105 cells) were seeded on a layer of previously polymerized Matrigel. The cell culture supernatants (1/10 volume of HUVEC media) obtained from HCT116 cells, left untreated or treated with quercetin (25 µM), were added to each well and incubated at 37°C. VEGF was used as a positive control. After 24 h, changes of cell morphology were captured through a phase-contrast microscope and photographed. The same experiment was done in the presence of a VEGF-neutralizing antibody (VEGF-Ab).

    Fig. 3. Quercetin up-regulates HIF-1 by inhibiting HIF prolyl hydroxylase. A, HCT116 cells were either left untreated or were pretreated with quercetin (50 µM) for 4 h, followed by the addition of cycloheximide for the indicated times. Levels of HIF-1 were visualized from nuclear extracts. Blots were reprobed for Topo II expression as a control for equivalent loading within each group. B, renal carcinoma cells that are deficient for VHL function (UMRC2) or a clonally selected line with VHL stably expressed (UMRC2/VHL) was treated with quercetin (50 µM) for 4 h, and HIF-1 protein was detected in nuclear extracts. C, a VHL capture assay using biotinylated HIF peptide was performed as described under Materials and Methods. Left, the assay was performed in the presence of cofactors and the indicated concentrations of quercetin, and resultant blots were probed for Flag (VHL). The control lane (con) represents the assay in the absence of added cofactors, whereas the untreated (UT) lane contains all required cofactors. Right, the same assay was repeated using a chemically hydroxylated peptide and increasing concentrations of quercetin. D, 293 cells, cotransfected with HA-HIF-1 and Flag-VHL, were treated with quercetin in the presence of MG-132 and were lysed 4 h later. HA-HIF-1 protein was immunoprecipitated by the addition of anti-HA antibody-bound beads. Immunoprecipitated proteins were solubilized in SDS sample buffer and separated by SDS-PAGE. Blots were probed with an anti-Flag antibody or an anti-HA antibody.

    Quercetin Up-Regulates HIF-1 by Inhibiting HIF Prolyl Hydroxylase. Because the -subunit of HIF is tightly regulated at the post-translational level by protein degradation (Huang et al., 1998), we considered whether quercetin modulated HIF-1 stability. HCT116 cells were either left untreated or were pretreated with quercetin for 4 h followed by the addition of the protein synthesis inhibitor cycloheximide for the indicated times and disappearance rate of HIF-1 protein was compared. As shown in Fig. 3A, HIF-1 protein was extremely labile, disappearing in 5 min in cells left untreated with quercetin. In marked contrast, substantial amount of HIF-1 protein still remained in quercetin-pretreated cells 40 min after the addition of cycloheximide, suggesting that quercetin stabilized HIF-1 protein. The central molecular mechanism for regulating HIF-1 protein stability is VHL-dependent proteasomal HIF degradation after hydroxylation of proline residues in HIF-1 by HPH (Ivan et al., 2001; Jaakkola et al., 2001). We considered whether quercetin might affect the HIF-regulating pathway. To test this, we first examined the effects of quercetin upon HIF-1 expression in either the parental VHL-deficient renal carcinoma cell line UMRC2, or UMRC2/VHL, which expresses a stably integrated construct encoding Flag-VHL (Isaacs et al., 2002). As shown in Fig. 3B, quercetin increased HIF-1 expression in UMRC2/VHL; however, when this experiment was repeated in the VHL-deficient parental line, quercetin was unable to induce HIF-1 expression. Although these data support an involvement of VHL in quercetin-mediated HIF induction, it remained unclear how quercetin intervenes in VHL-dependent HIF-1 regulation. Because HIF-prolyl hydroxylase is the key enzyme for VHL-dependent HIF degradation, we examined whether quercetin affected HIF-prolyl hydroxylase activity. To do this, we used an in vitro VHL capture assay (Bruick and McKnight, 2001; Isaacs et al., 2005) with a biotinylated HIF peptide that contains a conserved proline residue subject to HPH-dependent hydroxylation. As shown in Fig. 3C (left), the association of VHL with the HIF peptide in the absence of exogenously added cofactors (control lane) is undetectable. When the required cofactors for HPH are added (UT lane), the association between the HIF peptide and VHL is markedly enhanced. It is striking that a 25 µM concentration of quercetin significantly reduced the association between HIF and VHL, and a 100 µM concentration completely abrogated the interaction between these proteins. Finally, we used a chemically synthesized hydroxylated peptide to verify that quercetin directly affects HPH activity and does not impair VHL protein. As shown in Fig. 3C (right), quercetin does not impair the ability of VHL to associate with hydroxylated HIF peptide up to 200 µM. In contrast, an HIF peptide in which the two proline residues were mutated to alanine failed to bind VHL under any circumstances (data not shown). Our data strongly support the premise that quercetin is a potent inhibitor of HPH. To test this notion in cells, we transfected an HA-HIF-1 and a Flag-VHL plasmid in 293 cells followed by 4-h treatment with a proteasome inhibitor MG-132 in the presence or absence of quercetin. After immunoprecipitation with an anti-HA antibody, VHL levels in the immunocomplexes were monitored using an anti-Flag antibody. As shown in Fig. 3D, MG-132 increased the VHL level in the immunocomplex, and, consistent with the result of VHL capture assay, quercetin effectively prevented the increase of the VHL level.

    Two Chelating Moieties of Quercetin Are Involved in Inhibiting HPH. Our data demonstrate that quercetin inhibited HPH. We wished to explore how quercetin inhibited the enzyme. Because the enzyme requires cofactors to catalyze hydroxylation of HIF-1, we examined whether quercetin impaired the activity of the enzyme by affecting availability of the required factors. In vitro VHL capture assay was performed in the presence of various concentrations of the factors. As shown in Fig. 4A (top), whereas 10-fold increase of either 2-ketoglutarate or ascorbate did not at all affect the inhibitory effect of quercetin on the enzyme (data not shown), an escalating dose of iron attenuated the quercetin effect, as represented by restored VHL association. To further test this, cells were treated with quercetin in the presence of 200 µM iron, and HIF-1 level was monitored. Consistent with the in vitro result, HIF-1 induction was abolished in the iron-enriched condition (Fig. 4A, bottom). This result suggests that quercetin inhibits HPH activity by reducing the availability of iron. Quercetin reportedly is able to form metal complexes through three chelating moieties (Morel et al., 1994). To examine which chelating moiety of quercetin is required for inhibiting HPH, we treated cells with quercetin derivatives, namely 3',4'-hydroxyflavone, 3-hydroxyflavone, and 5-hydroxyflavone. As shown in Fig. 4B (top), 3',4'-hydroxyflavone and 3-hydroxyflavone but not 5-hydroxyflavone elevated the HIF-1 level. In addition, cells were treated with either 3',4'-hydroxyflavone or 3-hydroxyflavone in the iron-enriched condition. HIF-1 induction was abolished completely (Fig. 4B, middle), indicating that iron chelation via the moieties resulted in HIF-1 induction. To further test the requirement of the chelating ability for HIF-1 induction, cells were treated with the methylated forms of the quercetin derivatives with no chelating activity, and HIF-1 level was monitored. As shown in Fig. 4B (lower), the methylated derivatives failed to induce HIF-1. To ensure that the quercetin derivatives up-regulated HIF-1 via inhibiting HPH, VHL capture assay was carried out with the quercetin derivatives. As shown in Fig. 4C, as expected, 3',4'-hydroxyflavone decreased VHL association in a dose-dependent manner. It is interesting that although 3-hydroxyflavone increased the HIF-1 level in cells, it did not show an ability to attenuate VHL association, which represents no inhibition of HPH. We speculated that 3-hydroxyflavone might inhibit the enzyme by antagonizing interaction of the catalytic iron with ascorbate or 2-ketoglutarate in cells in which concentrations of them may be lower than those for the in vitro assay. In the process of HIF-1 hydroxylation by HPH, the transient association of the catalytic iron with ascorbate or 2-ketoglutarate is required (Kivirikko et al., 1989; Hewitson and Schofield, 2004). To test this possibility, we carried out the in vitro assay without exogenous addition of either ascorbate or 2-ketoglutarate. As reported previously (Ivan et al., 2002), HPH was still active and induced VHL association in the condition (data not shown). As shown in Fig. 4D, consistent with our hypothesis, 3-hydroxyflavone effectively reduced VHL association in the absence of exogenous ascorbate, and furthermore, the addition of ascorbate recovered VHL association. This phenomenon was not observed in the case of 2-ketoglutarate (data not shown). To further test the ascorbate effect, cells were treated with 3-hydroxyflavone in the presence of ascorbate or a cell-permeable ascorbate, and HIF-1 was monitored. As shown in Fig. 4E, pretreatment with ascorbate nullified the effect of 3-hydroxyflavone on HIF-1 induction.

    Fig. 4. Two chelating moieties of quercetin enable the flavonoid to inhibit HPH and subsequently induce HIF-1. A, top, VHL capture assay was performed in the presence of escalating dose of iron and quercetin (50 µM), and resultant blots were probed for Flag (VHL). Bottom, HCT cells were treated with quercetin (50 µM) in the presence or absence of various ferrous chloride, and HIF-1 protein levels were monitored in the nuclear extracts. B, top, HCT116 cells were treated with quercetin derivatives (50 µM), 3-hydroxyflavone (3-HF), 5-hydroxyflavone (5-HF), or 3',4'-dihydroxyflavone (DHF), and HIF-1 protein levels were monitored in the nuclear extracts. Middle, HCT116 cells were treated with 3-hydroxyflavone (50 µM) or 3',4'-dihydroxyflavone (50 µM) in the presence or absence of ferrous chloride (200 µM), and HIF-1 protein levels were monitored in the nuclear extracts. Bottom, HCT116 cells were treated with 3-hydroxyflavone (50 µM), 3',4'-dihydroxyflavone (50 µM), or their methylated forms, 3-methoxyflavone (100 µM, 3-MHF) or 3',4'-dimethoxyflavone (100 µM, DMF) and HIF-1 protein levels were monitored in the nuclear extracts. C, VHL capture assay was performed in the presence of 3-hydroxyflavone (50 µM), 3',4'-dihydroxyflavone (50 µM), and resultant blots were probed for Flag (VHL). D, VHL capture assay was performed with 3-hydroxyflavone (50 µM) in the presence or absence of ascorbate (100 µM), and resultant blots were probed for Flag (VHL). The reaction solution for this assay did not contain exogenous ascorbate. E, HCT116 cells were treated with 3-hydroxyflavone (50 µM) in the presence or absence of either ascorbate (V.C, 5 mM) or a cell-permeable ascorbate, (+)-5,6-O-Isopropylidene-L-ascorbic acid (Iso V.C, 1 mM), and HIF-1 protein levels were monitored in the nuclear extracts.

    In this study, we demonstrate that quercetin, the aglycone of rutin, induced an ulcer-healing factor, VEGF, via activating the HIF-1 pathway most likely by inhibiting HIF-prolyl hydroxylase. Furthermore, we provide information on the structural requirement of quercetin for inhibiting the enzyme.

    Quercetin was reported to stabilize HIF-1 protein and induce VEGF in endothelial cells (Wilson and Poellinger, 2002). In line with the report, we found that quercetin showed the same effect in colon epithelial cells and the inflamed colonic tissue. Moreover, we clarified HIF-1 dependence in quercetin-mediated VEGF induction by showing that quercetin was able to induce VEGF in normal mouse hepatoma cells but not in the matched cells with genetically defective HIF-1 and consequent inactivation of HIF-1 pathway. Furthermore, we elucidated the molecular mechanism underlying quercetin-mediated HIF-1 up-regulation and subsequent activation of HIF-1 pathway. Our data showing that quercetin delayed the degradation of HIF-1 protein and quercetin-mediated HIF-1 induction occurred only in cells with functional VHL suggest that the flavonoid stabilizes HIF-1 protein by preventing VHL-dependent HIF-1 degradation. This hypothesis is validated by providing compelling evidence that quercetin inhibited HPH, thus interfering with the hydroxylation of HIF-1, a critical post-translational modification for VHL-dependent degradation of HIF-1. In the in vitro VHL capture assay, in which in vitro-translated HPH hydroxylates its substrate, HIF-peptide, and subsequently the hydroxylated peptide associates with in vitro-translated VHL, we observed that the addition of quercetin before reaction of HPH with HIF peptide attenuated VHL association with HIF peptide, indicating reduced hydroxylation of HIF peptide by quercetin-mediated HPH inhibition. This in vitro result was confirmed by demonstrating that quercetin decreased the level of VHL precipitated together with HIF-1 in cells. We suggest that HPH inhibition by quercetin occurred via reducing the availability of iron, an essential cofactor of the enzyme. This argument was supported by the data showing that 1) escalation of iron dose attenuated the quercetin effect on VHL association; 2) pretreatment with iron prevented HIF-1 protein induction by quercetin; and 3) dose changes of the other factors, ascorbate and 2-ketoglutarate, did not influence the quercetin effects on VHL association. In line with this finding, it was revealed that chelating moieties of quercetin were involved in HPH inhibition. Although quercetin (3,5,7,3'4'-pentahydroxyflavone) has three chelating moieties in it, two of them seem to be used to reduce the availability of iron, as demonstrated in the data showing that 3',4'-dihydroxyflavone and 3-hydroxyflavone but not 5-hydroxyflavone decreased VHL association with HIF-1 peptide and elevated HIF-1 level. Furthermore, the methylated forms of 3',4'-dihydroxyflavone and 3-hydroxyflavone, which are not able to form an iron complex, lost the ability to decrease the VHL association (Supplemental Data 1A) and elevate HIF-1 level; furthermore, pretreatment with iron neutralized the effects of the two flavones, 3',4'-dihydroxyflavone and 3-hydroxyflavone (Supplemental data 1B). Because, for hydroxylation of the substrate HIF-1 by HIF prolyl hydroxylase, iron in the catalytic site needs to be associated transiently with the required factors, 2-ketoglutarate and ascorbate (Kivirikko et al., 1989; Hewitson and Schofield, 2004), it is likely that the iron chelation with the flavones prevents the association, thus resulting in impairing the catalytic activity of HPH. This hypothesis is supported by our observation that 3-hydroxyflavone inhibition of HPH and 3-hydroxyflavone induction of HIF-1 were abolished completely by the addition of ascorbate. In the case of 3',4'-dihydroxyflavone, the HPH inhibition was not recovered in the presence of excess amount of 2-ketoglutarate or ascorbate (Supplemental Data 1C), suggesting that the affinity between the flavone and iron is too strong to be replaced with the required factors, or the HPH inhibition occurs by a way regardless of antagonizing the required factors. Although it is not clear which of the two chelating moieties of quercetin associates preferentially with the iron in the enzyme catalytic site, 3',4'-dihydroxyl group should be the preferential partner for the iron chelation in the condition in which ascorbate exists over the concentration for the antagonism. In fact, the observation that quercetin still inhibits HPH in the presence of 1 mM ascorbate manifests the situation.

    In our previous report, we suggested that quercetin-mediated amelioration of rat colitis was elicited by inhibiting NFB activity (Kim et al., 2005). Our data demonstrating quercetin activation of HIF-1-VEGF pathway in cells and inflamed colonic tissue suggest that quercetin exerts its clinical effect by not only inhibiting the inflammatory signal but also activating the tissue repair signal. Although the activation of HIF-1-VEGF pathway promotes angiogenesis that could enhance repair of injured mucosa, quercetin per se was reported to inhibit growth of endothelial cells (Tan et al., 2003), thus possessing antiangiogenic potential at the concentration range used in this experiment. However, it is more likely that quercetin exerted a positive effect on angiogenesis in the colitis rat model. This is based on the following reasons: 1) unlike the previous report in which endothelial cells were exposed continually to quercetin for 1 to 3 days, quercetin in the large intestine remained at 200 to 30 µM for approximately 6 h after oral administration of a glycoside of quercetin (Kim et al., 2005), which may be enough for quercetin to activate HIF-1-VHL pathway but not to inhibit growth of endothelial cells; 2) because concentration of quercetin diffused to the colonic endothelial cells should be lower than that of quercetin in the gut lumen; and 3) considering the second reason stated above, the tube formation assay, in which the diluted supernatant of quercetin-treated cells was used, may reflect the action of quercetin in the inflamed large intestine and, indeed, VEGF in the diluted supernatant promoted tube formation of HUVEC cells.

    Although the exact molecular mechanism by which quercetin inhibits growth of endothelial cells is not known yet, a number of reports suggest that quercetin elicits the cytotoxic effect by blocking cell proliferation signals such as phosphatidylinositol-3 kinase (Agullo et al., 1997; Williams et al., 2004). According to recent reports on the relationship between flavonoid structure and either phosphatidylinositol-3 kinase or NFB (Agullo et al., 1997; Chen et al., 2004), whereas the double bond (C2–C3) of the C ring is important for both biological activities, the effect of hydroxyl moiety (-OH) of flavonoids on the protein targets varies depending on position and number of the moiety on the flavonoid skeleton. In addition to this structural information, our data on the structural requirement of quercetin for HPH inhibition may provide a possibility to separate antiangiogenic activity from biological activities of quercetin (for amelioration of IBD) via structural modification. Moreover, our data add one to the structural criteria that could be used to predict biological activities of flavonoid.

    Acknowledgements

    We thank Dr. Jung-Ae Kim (Yeungnam University) for quantitative analysis of tuber formation.

    ABBREVIATIONS: IBD, inflammatory bowel disease; HIF-1, hypoxia inducible factor-1; NFB, nuclear factor-B; HPH, hypoxia inducible factor-proly hydroxylase; VEGF, vascular endothelial growth factor; TNBS, 2,4,6-trinotrobenzene-sulfonic acid; VHL, von Hippel Lindau; ARNT, aryl hydrocarbon receptor nuclear translocator; PMSF, phenylmethylsulfonyl fluoride; HA, hemagglutinin; PAGE, polyacrylamide gel electrophoresis; IVT, in vitro-translated; ELISA, enzyme-linked immunosorbent assay; MG-132, N-benzoyloxycarbonyl (Z)-Leu-Leu-leucinal.

    The online version of this article (available at http://molpharm.aspetjournals.org) contains supplemental material.

【参考文献】
  Agullo G, Gamet-Payrastre L, Manenti S, Viala C, Remesy C, Chap H, and Payrastre B (1997) Relationship between flavonoid structure and inhibition of phosphatidylinositol 3-kinase: a comparison with tyrosine kinase and protein kinase C inhibition. Biochem Pharmacol 53: 1649–1657.[CrossRef][Medline]

Andrews NC and Faller DV (1991) A rapid micropreparation technique for extraction of DNA-binding proteins from limiting numbers of mammalian cells. Nucleic Acids Res 19: 249919.

Baatar D, Jones MK, Tsugawa K, Pai R, Moon WS, Koh GY, Kim I, Kitano S, and Tarnawski AS (2002) Esophageal ulceration triggers expression of hypoxia-inducible factor-1 alpha and activates vascular endothelial growth factor gene: implications for angiogenesis and ulcer healing. Am J Pathol 161: 1449–1457.[Abstract/Free Full Text]

Beck PL and Podolsky DK (1999) Growth factors in inflammatory bowel disease. Inflamm Bowel Dis 5: 44–60.[Medline]

Bruick RK and McKnight SL (2001) A conserved family of prolyl-4-hydroxylases that modify HIF. Science (Wash DC) 294: 1337–1340.[Abstract/Free Full Text]

Chen CC, Chow MP, Huang WC, Lin YC, and Chang YJ (2004) Flavonoids inhibit tumor necrosis factor-alpha-induced up-regulation of intercellular adhesion molecule-1 (ICAM-1) in respiratory epithelial cells through activator protein-1 and nuclear factor-B: structure-activity relationships. Mol Pharmacol 66: 683–693.[Abstract/Free Full Text]

Cockman ME, Masson N, Mole DR, Jaakkola P, Chang GW, Clifford SC, Maher ER, Pugh CW, Ratcliffe PJ, and Maxwell PH (2000) Hypoxia inducible factor- binding and ubiquitylation by the von Hippel-Lindau tumor suppressor protein. J Biol Chem 275: 25733–25741.[Abstract/Free Full Text]

Hashimoto H, Akimoto M, Maeda A, Shigemoto M, and Yamashito K (2004) Relation of hypoxia-inducible factor-1alpha to vascular endothelial growth factor and vasoactive factors during healing of gastric ulcers. J Cardiovasc Pharmacol 44: S407–S409.[CrossRef][Medline]

Hewitson KS and Schofield CJ (2004) The HIF pathway as a therapeutic target. Drug Discov Today 9: 704–711.[CrossRef][Medline]

Huang LE, Gu J, Schau M, and Bunn HF (1998) Regulation of hypoxia-inducible factor 1 is mediated by an O2-dependent degradation domain via the ubiquitin-proteasome pathway. Proc Natl Acad Sci USA 95: 7987–7992.[Abstract/Free Full Text]

Isaacs JS, Jung YJ, Mimnaugh EG, Martinez A, Cuttitta F, and Neckers LM (2002) Hsp90 regulates a von Hippel Lindau-independent hypoxia-inducible factor-1 -degradative pathway. J Biol Chem 277: 29936–29944.[Abstract/Free Full Text]

Isaacs JS, Jung YJ, Mole DR, Lee S, Torres-Cabala C, Chung YL, Merino M, Trepel J, Zbar B, Toro J, et al. (2005) HIF overexpression correlates with biallelic loss of fumarate hydratase in renal cancer: novel role of fumarate in regulation of HIF stability. Cancer Cell 8: 143–153.[CrossRef][Medline]

Ivan M, Haberberger T, Gervasi DC, Michelson KS, Gunzler V, Kondo K, Yang H, Sorokina I, Conaway RC, Conaway JW Jr, et al. (2002) Biochemical purification and pharmacological inhibition of a mammalian prolyl hydroxylase acting on hypoxia-inducible factor. Proc Natl Acad Sci USA 99: 13459–13464.[Abstract/Free Full Text]

Ivan M, Kondo K, Yang H, Kim W, Valiando J, Ohh M, Salic A, Asara JM, Lane WS, and Kaelin WG Jr (2001) HIFalpha targeted for VHL-mediated destruction by proline hydroxylation: implications for O2 sensing. Science (Wash DC) 292: 464–468.[Abstract/Free Full Text]

Jaakkola P, Mole DR, Tian YM, Wilson MI, Gielbert J, Gaskell SJ, Kriegsheim A, Hebestreit HF, Mukherji M, Schofield CJ, et al. (2001) Targeting of HIF-alpha to the von Hippel-Lindau ubiquitylation complex by O2-regulated prolyl hydroxylation. Science (Wash DC) 292: 468–472.[Abstract/Free Full Text]

Kim H, Kong H, Choi B, Yang Y, Kim Y, Lim MJ, Neckers L, and Jung Y (2005) Metabolic and pharmacological properties of rutin, a dietary quercetin glycoside, for treatment of inflammatory bowel disease. Pharm Res (NY) 22: 1499–1509.

Kivirikko KI, Myllyla R, and Pihlajaniemi T (1989) Protein hydroxylation: prolyl 4-hydroxylase, an enzyme with four cosubstrates and a multifunctional subunit. FASEB J 3: 1609–1617.[Abstract]

Li H, Ko HP, and Whitlock JP (1996) Induction of phosphoglycerate kinase 1 gene expression by hypoxia. Roles of Arnt and HIF1 . J Biol Chem 271: 21262–21267.[Abstract/Free Full Text]

Mammen JM and Matthews JB (2003) Mucosal repair in the gastrointestinal tract. Crit Care Med 31: S532–S537.[CrossRef][Medline]

Morel I, Lescoat G, Cillard P, and Cillard J (1994) Role of flavonoids and iron chelation in antioxidant action. Methods Enzymol 234: 437–443.[Medline]

Morris GP, Beck PL, Herridge MS, Depew WT, Szewczuk MR, and Wallace JL (1989) Hapten-induced model of chronic inflammation and ulceration in the rat colon. Gastroenterology 96: 795–803.[Medline]

Nijveldt RJ, van Nood E, van Hoorn DE, Boelens PG, van Norren K, and van Leeuwen PA (2001) Flavonoids: a review of probable mechanisms of action and potential applications. Am J Clin Nutr 74: 418–425.[Abstract/Free Full Text]

Salceda S and Caro J (1997) Hypoxia-inducible factor 1 (HIF-1) protein is rapidly degraded by the ubiquitin-proteasome system under normoxic conditions. Its stabilization by hypoxia depends on redox-induced changes. J Biol Chem 272: 22642–22647.[Abstract/Free Full Text]

Semenza GL (1998) Hypoxia-inducible factor 1: master regulator of O2 homeostasis. Curr Opin Genet Dev 8: 588–594.[CrossRef][Medline]

Tan WF, Lin LP, Li MH, Zhang YX, Tong YG, Xiao D, and Ding J (2003) Quercetin, a dietary-derived flavonoid, possesses antiangiogenic potential. Eur J Pharmacol 459: 255–262.[CrossRef][Medline]

Tanimoto K, Makino Y, Pereira T, and Poellinger L (2000) Mechanism of regulation of the hypoxia-inducible factor-1 alpha by the von Hippel-Lindau tumor suppressor protein. EMBO (Eur Mol Biol Organ) J 19: 4298–4309.[CrossRef][Medline]

Tarnawski A (2006) Gene therapy for ulcerations of the gastrointestinal tract. Drugs Today (Barc) 42: 193–203.[CrossRef][Medline]

Tarnawski AS (2005) Cellular and molecular mechanisms of gastrointestinal ulcer healing. Dig Dis Sci 50 (Suppl 1): S24–S33.[CrossRef][Medline]

Wang GL, Jiang BH, Rue EA, and Semenza GL (1995) Hypoxia-inducible factor 1 is a basic-helix-loop-helix-PAS heterodimer regulated by cellular O2 tension. Proc Natl Acad Sci USA 92: 5510–5514.[Abstract/Free Full Text]

Wiener CM, Booth G, and Semenza GL (1996) In vivo expression of mRNAs encoding hypoxia-inducible factor 1. Biochem Biophys Res Commun 225: 485–488.[CrossRef][Medline]

Williams RJ, Spencer JP, and Rice-Evans C (2004) Flavonoids: antioxidants or signalling molecules. Free Radic Biol Med 36: 838–849.[CrossRef][Medline]

Wilson WJ and Poellinger L (2002) The dietary flavonoid quercetin modulates HIF-1 alpha activity in endothelial cells. Biochem Biophys Res Commun 293: 446–450.[CrossRef][Medline]

Yano H, Hirayama F, Kamada M, Arima H, and Uekama K (2002) Colon-specific delivery of prednisolone-appended alpha-cyclodextrin conjugate: alleviation of systemic side effect after oral administration. J Control Release 79: 103–112.[CrossRef][Medline]

作者单位：Laboratory of Biomedicinal (H.J., H.K., D.C., D.K., Y.J.)/Medicinal Chemistry (Y.M.K.), College of Pharmacy, Pusan National University, Busan, Korea; and Department of Molecular Biology, College of Nature Sciences, Pusan National University, Busan, Korea (S.Y.P., Y.J.K.)