Address correspondence and reprint requests to Douglas L. Feinstein, Department of Anesthesiology, University of Illinois, 1819 West Polk Street, Room 544, Chicago, IL 60612, USA. E-mail: firstname.lastname@example.org
The expression of inducible nitric oxide synthase (NOS2) in glial cells is inhibited by neurotransmitters such as norepinephrine (NE) which elevate cAMP levels. We examined the molecular basis for this effect using a 2.2-kb fragment of the rat NOS2 promoter transfected into rat C6 glioma cells. Promoter activation (up to six-fold) by lipopolysaccharide (LPS) and interferon-γ (IFNγ) was reduced by NE, which alone had no effect. However, a promoter construct extending to bp −130 and containing the proximal nuclear factor-kappa B (NF-κB) binding site was minimally activated by LPS and cytokines, but activated up to three-fold by NE. Deletion analysis identified a 27-bp region (bp −187 to −160) as critical for mediating this suppressive effect. This region also enhanced promoter activation by LPS and cytokines, and prevented activation by NE alone. Gel shift analysis revealed constitutive binding to this region, and induction by NE of additional complexes which could be blocked by an antibody against CREB. NE also increased levels of the IκBα protein which could contribute to its suppressive effects. These results identify a critical role for this 27-bp region in regulation of NOS2 promoter activation and suppression by cAMP.
the calcium independent isoform of nitric oxide synthase, also referred to as iNOS
rat aortic smooth muscle cells
sodium dodecyl sulfate–polyacrylamide gel electrophoresis
tumour necrosis factor-α
rat vein smooth muscle cells.
The production of nitric oxide (NO) in brain is catalyzed by members of the nitric oxide synthase (NOS) family. Constitutively expressed NOS1 and NOS3 synthesize NO in a tightly regulated manner, and their activities are calcium and calmodulin dependent. NO derived from these enzymes is involved in neural signaling and vasodilation. The NOS2 isotype (also referred to as inflammatory NOS, or iNOS) is normally not present, but its expression is induced as a consequence of inflammatory stimuli. NO derived from NOS2 is often involved in cytotoxic and cytostatic functions, contributing to the destruction of invading organisms and tumor cells (MacMicking et al. 1997). Studies using primary astrocyte cultures (Galea et al. 1992; Simmons and Murphy 1992; Hewett et al. 1993; Lee et al. 1993) and C6 astroglioma cells (Feinstein et al. 1994; Simmons and Murphy 1993) demonstrate that NOS2 can be induced in astrocytes by stimulation with bacterial endotoxin lipopolysaccharide (LPS) or a combination of pro-inflammatory cytokines. In brain, astroglial NOS2 expression has been described in several diseases including human multiple sclerosis (Bo et al. 1994; Bagasra et al. 1995), after cerebral ischemia (Endoh et al. 1994), in AIDS dementia (Hori et al. 1999), and in Alzheimer's disease (Vodovotz et al. 1996; Wallace et al. 1997). It is generally considered that the unregulated activity of NOS2 contributes to the development, or damage occurring in these diseases.
We previously demonstrated that incubation of glial cells with norepinephrine (NE) reduces NOS2 induction, mediated by binding to beta adrenergic receptors (β-ARs), increase in intracellular cAMP, and activation of protein kinase A (Feinstein et al. 1993; Feinstein 1998). Suppression of glial cell NOS2 expression by other agents which increase cAMP has also been described (Oda et al. 1997; Pahan et al. 1997), although using a mouse astrocyte cell line the cAMP mimetic 8-bromo-cAMP potentiated expression (Burgher et al. 1997). cAMP has also been shown to suppress NOS2 expression in other cell types, including Kupffer cells, macrophages, hepatocytes, and islet cells (see Galea and Feinstein 1999 for review). In contrast, studies using peritoneal cells, smooth muscle cells, mesangial cells, macrophages, and adipocytes indicate that cAMP can potentiate cytokine-dependent NOS2 expression; and in some cases cAMP alone was able to induce NOS2 expression. Whether increased cAMP acts as an inhibitor vs. a stimulator of NOS2 expression may therefore depend upon the particular cell type used, the agents used to increase cAMP, the duration and magnitude of the increase in cAMP, and the methods used to assess NOS2 expression. However, the observations that NOS2 expression in smooth muscle cells and mesangial cells is reproducibly increased by cAMP, while in glial cells, hepatocytes, and islet cells it is generally reduced, suggests the existence of cell specific suppressors and/or activators of NOS2 expression.
To elucidate the molecular mechanisms underlying regulation of NOS2 expression and modulation by NE, we carried out studies using portions of the mouse NOS2 promoter (Xie et al. 1993) expressed in rat C6 glioma cells (Feinstein 1998). We demonstrated that NE reduced transcriptional activation by LPS and cytokines of a 1588-bp promoter construct, but not of a minimal 88-bp construct which extends only to the 5′ end of the proximal nuclear factor-kappa B (NF-κB) binding site. We also showed that whereas NE did not activate the 1588-bp promoter, it did activate the smaller, 88-bp promoter. These results confirmed that the inhibitory effects of NE are mediated by regulation of transcription factor binding to the NOS2 promoter, and further suggested that stimulatory effects of NE (and presumably cAMP) are mediated at a region located near the transcriptional start site.
To further define the trans-acting factors and cis-elements responsible for regulation of NOS2 expression by NE, we used a 2.2-kb portion of the rat NOS2 promoter for studies in rat glioma cells. Our current results define a small (27-bp) region of the rat NOS2 promoter which confers both cytokine-dependent activation and its suppression by NE. Furthermore, the presence of this region restricts promoter activation by NE alone. These studies should help define the cell specific factors which dictate whether cAMP acts as an inhibitory vs. a stimulatory regulator of inflammatory gene expression.
Materials and methods
Cell culture reagents (DMEM, antibiotics and LPS (Salmonella typhirium) were from Sigma (St Louis, MO, USA). Fetal calf serum (FCS) was from Atlanta Biological (Norcross, GA, USA). Human recombinant interleukin 1β (IL-1β; 4 × 106 unit/mg) was obtained from the NIH AIDs reagents program. Recombinant rat interferon γ (IFN-γ; 4 × 106 unit/mg), synthetic oligonucleotides, and geneticin were from GIBCO (Gaithersburg, MD, USA). Anti-NF-κB p50 (SC-115) and p65 (SC-109) subunits, CEBP (SC-746x, reactive against CEBPα, β and δ), CREB (SC-186), and Oct-1 (SC-232x) rabbit polyclonal antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Taq polymerase and cDNA reagents were from Promega (Madison, WI, USA) and GIBCO.
Cloning of the rat NOS2 promoter
A 2168-bp fragment of the NOS2 promoter was amplified by PCR from isolated rat liver genomic DNA using synthetic oligonucleotide primers derived from a published rat NOS2 promoter sequence (Zhang et al. 1998; GenBank accession number AF042085): RatNOSg1, 5′-CAGCCAAGTATTCCAAAGCAA-3′ (bases 1108–1127); and RatNOSg2, 5′-AGTCCAGTCCCCTCAC CAA-3′ (bases 3259–3277). The resulting PCR product was eluted from low-melting point agarose, purified, and subcloned into the pGEM-T vector to yield pGEMT-NOS2g. Colonies were screened by PCR for the presence of the NOS2 insert, clone orientation determined by restriction enzyme analysis of purified plasmid DNA, and two clones having opposite insert orientations selected for automated DNA sequence analysis. The DNA sequence contained two differences vs. the published NOS2 promoter sequence (bp 1660, C for T; and bp 1767, G for A) thus confirming its identity as the rat NOS2 promoter. This fragment contains 2168 bp of DNA, begins at bp −2088 relative to the transcriptional start site and extends to bp + 80.
The insert from pGEMT-NOS2g was excised with SacI and SacII restriction enzymes and cloned into the same sites of pBluescript II KS(+) (Stratagene, CA, USA). Resulting plasmid pBL-rNOS2 was digested with SacI and SmaI and subcloned into the same sites in vector pGL3-Basic (Promega, WI, USA) containing the luciferase reporter gene to yield pGL3–2.2. Deletions from the 5′ end of the NOS2 promoter were made by PCR amplification of pBL-rNOS2 using internal NOS2 promoter primers having an Xho1 site at the 5′ end, and a universal T3 primer present in the KS plasmid, digesting the PCR product with Xho1, and cloning into Xho1-digested pGL3-basic. A schematic of these plasmids is shown in Fig. 1(a).
C6 glioma cells were grown in Dulbecco's modified Eagle's medium (DMEM) containing 10% FCS and antibiotics (penicillin and streptomycin). Cells were passaged once a week, and used after 3–4 days at which point they were 90–95% confluent. Primary astrocytes were prepared from cerebral cortices of post natal day 1 Sprague–Dawley rats as previously described (Galea et al. 1992). Media was changed every 3 days. After 2 weeks of growth in complete media (DMEM with 10% FCS) the cultures consisted of 95–98% astrocytes and 2–3% microglia.
Production of stably transfected rat C6 cell lines
As in our previous studies (Feinstein 1998; Stasiolek et al. 2000), we used stable transfectants for the current studies, which provides essentially 100% transfection efficiency and thus obviates the need for cotransfection with a second reporter gene to normalize for differences in transfection efficiencies. This also allows us to grow cells to confluency or near confluency cells which results in higher levels of NOS2 induction. Rat C6 glioma cells were grown in Ham's media containing 5% FCS, 20 mm HEPES and antibiotics until 40% confluent. Cells were transfected with rat NOS2 promoter constructs in the pGL3 vector (or pGL3 vector alone) using lipofectamine (GIBCO) according to manufacturer's recommendations. The C6 cells were cotransfected with the pSV2Neo vector (Stratagene), containing the neomycin resistance gene. After 1.5 day of incubation with normal medium (Ham's, 10% FCS, antibiotics) stable transfectants were selected by growth in 0.8 mg/mL of antibiotic G418.
NOS2 induction and activity measurements
Unless indicated, cells were grown to 80–90% confluency, the growth media removed, the cells washed in serum-free media, and incubations carried out in fresh DMEM containing antibiotics and 1% FCS. NOS2 was induced in C6 cells and transfected cell lines by incubation with bacterial endotoxin lipopolysaccharide (LPS, 1 µg/mL), recombinant rat IFN-γ (20 U/mL), recombinant human IL-1β (10 ng/mL) or various combinations for indicated times. The induction of endogenous NOS2 was assessed indirectly by nitrite production in the cell culture media after 18–24 h. An aliquot of the cell culture media (80–100 µL was mixed with one-half volume of Griess' reagent (Green et al. 1982) and the absorption measured at 550 nm in a microplate reader. Solutions of NaNO2 dissolved in DMEM/1% FCS served as standards, and the absorption due to cells incubated in media alone (no NOS2 inducers) was subtracted from experimental values.
Luciferase activity assay
After indicated incubation times, cells were washed twice in ice-cold PBS. To prepare lysates, 50 µL of CHAPS buffer (10 mm CHAPS, 10 mm Tris pH 7.4.) were added to each well, the plate frozen at −80°C, thawed, and shaken on a rotary shaker for 10–15 min at room temperature (22°C). Aliquots of cell lysates (10–20 µL) containing equal amounts of protein (10–20 µg) were placed into wells of an opaque, white 96-well microplate. An equal volume of luciferase substrate (Steady Glo reagent, Promega) was added to all samples, and the luminescence measured in a microplate luminometer (Rosys-Anthos, Anthos, Durham, NC, USA).
Preparation of cell extracts and EMSA
Following stimulation with NOS2 inducers, cells were washed in cold PBS, pelleted, and resuspended in hypotonic buffer (10 mm HEPES pH 7.6, 1 mm EDTA, 10 mm KCl, 1 mm DTT) and protease and phosphatase inhibitors (10 µg/mL aprotinin and leupeptin, 100 µm TPCK, 1 mm PMSF, 30 mmβ-glycerophospate, 50 mm NaF, 1 mm Na3VO4 and 20 mm pNPP). After 15 min on ice, Nonidet P-40 was added to a final concentration of 0.6%, the lysates incubated a further 5 min, and then centrifuged for 15 min at 12 000 g to pellet nuclei. The nuclei were washed once in the same buffer by gentle resuspension and centrifugation. Nuclear extracts were prepared by extraction of the nuclear pellet for 15 min at 4°C in 50–100 µL of 400 mm NaCl; 10 mm HEPES pH 7.9; 1 mm EDTA; 1 mm dithiothreitol (DTT); and 1 mm phenylmethylsulfonyl fluoride (PMSF). One to two micrograms of nuclear extract were incubated for 30 min at room temperature in 10 µL containing 50 mm NaCl; 200 ng poly dI:dC; 1 mm DTT; 5% glycerol; 10 mm HEPES pH 7.9; 1 mm PMSF; 50 µg/mL aprotinin and 100 000 dpm (0.5 ng) of double-stranded oligonucleotide radiolabeled with [32P]ATP using T4 kinase. Supershift EMSAs were carried out by pre-incubating nuclear extracts with 1 µg of specific antibody at room temperature for 20 min before addition of probe. The probes derived from the rat NOS2 promoter are indicated in Fig. 1(b) and were: (1) 22-bp oligo1 (bp −114 to −93) containing the proximal NF-κB binding site; (2) 58-bp oligo2 (bp −187 to −130); and (3) 21-bp oligo3 (bp −178 to −158). Mutant-oligo3 has three differences from oligo3 (marked in bold: 5′-ACAGAGTGATGTAATCATCCA-3′) and therefore lacks the consensus CRE site but maintains the consensus C/EBP site. The reaction products were separated by electrophoresis at 4°C for 2.5 h at 100 V through 4.5% acrylamide gels containing 0.5X Tris-borate-EDTA (TBE) buffer which had been pre-run for 1 h at 100 V, dried, and exposed to X-ray film for up to 3 days.
Total cytoplasmic RNA was prepared from cells by the NP40-lysis procedure and mRNA levels were estimated by competitive RT-PCR assay (Galea et al. 1994a). The primers used for NOS2 detection were 1704F (5′-CTGCATGGAACAGTATAAGGCA AAC-3′), corresponding to bases 1704–1728; and 1933R (5′-CAGACAGTTTCTGGTCGATGTCATGA-3′), complementary to bases 1908–33 of rat iNOS cDNA sequence (Galea et al. 1994) which yield a 230-bp product. The primers used for IκBα were 299F (5′-CATGAAGAGAAGACACTGAC CATGGAA-3′) and 627R (5′-TGGATAGAGGCTAAGTGTAGA CACG-3′) which yield a 328-bp product. The primers used for glyceraldehyde 3-phosphate dehydrogenase (GDH) detection were 796F (5′-GCCAAGTATGATGACATCAAGAAG) and 1059R (5′-TCCA GGGGTTTCTTACTCCTTGGA) which yield a 264-bp product. PCRs were peformed in the presence of a known amount of a smaller (by 50 bp) competitive internal standard (CIS) which is amplified with the same efficiency as the cDNA template. PCR was initiated by a hot start method, and conditions were 35 cycles of denaturation at 93°C for 35 s; annealing at 63°C for 45 s; and extension at 72°C for 45 s; followed by 10 min at 72°C in a Hybaid Thermoreactor (Franklin, MA, USA) controlled by tube temperature. PCR products were separated by electrophoresis through 2% agarose gels containing 0.1 µg/mL ethidium bromide. Band intensities were determined using the Alpha Infotech 2000 imaging system.
Preparation of protein lysates and western blot analysis
Cells were washed twice with ice-cold PBS, scraped from the dishes, and collected by centrifugation (3000 g for 5 min). The cells were resuspended directly into five volumes of 8 m urea. Aliquots were mixed with an equal volume of 2 × buffer containing 124 mm Tris-Cl pH 6.8, 0.2% sodium dodecyl sulfate (SDS), 10% beta mercapto ethanol, 10 mm EDTA, 50% glycerol and boiled for 5 min. Protein samples (approximately 20 µg) were separated through 10% SDS–polyacrylamide gel electrophoresis (SDS–PAGE), then transferred to PVDF membranes by semi-dry electrophoretic transfer. The membranes were blocked in Tris-buffered saline containing 0.1% Tween-20 (TBST) and 5% dry milk (1 h), rinsed, and incubated with primary antibodies in TBST containing 0.5% BSA overnight with gentle shaking at 4°C. Membranes were washed four times in TBST, and 0.1 µg/mL peroxidase-labeled goat secondary antibodies added for 1 h. Following four washes in TBST, bands were visualized by incubation in enhanced chemiluminescence reagents, and exposure to X-ray film. Polyclonal antibodies directed against IκBα (SC-371, which recognizes the carboxy terminus of IκBα) and IκBβ (SC-945, which recognizes the carboxy terminus of the 43 kDa IκBβ isoform) were used at 1 : 1500 dilution and were from Santa Cruz Biotechnology.
All enzymatic experiments were performed at least in triplicate, and means ± SEM determined. EMSAs were peformed at least in duplicate using independent preparations of nuclear extracts. Statistical significance was assessed using Prism 3.0 software (GraphPad, San Diego, CA, USA) by one way anova analysis followed by Dunnett's multiple comparison, and two-way anova followed by Bonferroni's post hoc tests, and p-values < 0.05 were considered significant.
Cloning of rat NOS2 promoter constructs
A 2.2-kb fragment of the rat NOS2 promoter was obtained by PCR amplification of Sprague–Dawley rat liver genomic DNA using primers designed according to the published sequence (Zhang et al. 1998). Plasmid pBL-NOS2 was used as a template to produce 5′ end deletion constructs which were subcloned into pGL3-basic luciferase vector (Fig. 1a); the first (pGL3-CREB) beginning at bp −187, the second (pGL3-ΔCREB) at bp −160, and the third (pGL3-κB) at bp −130 relative to the transcriptional start site, and all extending to bp + 80. Plasmid pGL3-κB contains consensus binding sites for transcription factors NF-κB and C/EBP. Plasmid pGL3-CREB contains an additional 57 bp compared with pGL3-κB and additional binding sites for CREB and a second C/EBP site. Plasmid pGL3-ΔCREB lacks the 27 bp present at the 5′ end of pGL3-CREB and therefore lacks the consensus bindings sites for CREB and C/EBP. These plasmids were used to produce stable transfectants in rat C6 cells, thus providing us with a homogenous, reproducible source of 100% transfected C6 cells. The promoterless pGL3-basic vector in C6-basic cells exhibited a low basal promoter activity (0.1–0.5 RLUs per second per µg protein). Basal promoter activity was similar in C6-CREB and C6–2.2 cells and was several hundred fold higher than in C6-basic cells (40–80 RLUs per second per µg protein). Basal promoter activity of C6-ΔCREB and C6-κB cells was 10–15 times lower (3–6 RLUs per second per µg protein).
Effect of incubation time on promoter activation
The magnitude of NOS2 promoter induction depended upon the incubation time with inducers (Fig. 2a). For both C6–2.2 and C6-CREB cells, luciferase activity was observed as soon as 3 h incubation, activity continued to increase up to 8 h, and decreased by 16 h. The maximal level of activation in both C6–2.2 and C6-CREB cells was similar, and was between 3.5 and 7-fold vs. non-stimulated cells. In both C6–2.2 and C6-CREB cells, the luciferase activity was statistically different at 4 and 8 h vs. luciferase activity measured in non-stimulated cells. In contrast, in C6-κB cells the maximal increase of promoter activation due to LPS and IFNγ was only 20–30% above control values at any of the time points measured, which was not statistically different vs. non-stimulated cells.
These experiments point to a variability in NOS2 promoter activation dependent upon time of incubation with inducers, and in other experiments (data not shown) the starting cell density and the number of media changes made during cell growth. For these reasons, subsequent promoter assays were routinely peformed using cells initially plated out at 60% level of confluency, grown for 3 days without any media change and incubated for 6–8 h with inducers.
Effects of NE on NOS2 promoter activation
As previously shown (Feinstein et al. 1993; Feinstein 1998), incubation of C6 cells with LPS plus IFNγ, or primary astrocytes with LPS alone, induced accumulation of nitrites in the cell culture media and cytosolic NOS2 mRNA both which were decreased by co-incubation with NE (Fig. 3). We previously showed that activation of a 1588-bp mouse NOS2 promoter by LPS plus IFNγ was also reduced (by approximately 50%) by co-incubation with NE (Feinstein 1998). NE also reduced the LPS plus IFNγ dependent promoter activation of the 2.2-kb rat NOS2 promoter in C6–2.2 cells (Fig. 2b) but to a lesser extent than that observed using the mouse promoter. In contrast, NE significantly reduced promoter activation by LPS and IFNγ in C6-CREB cells. In C6-κB cells, NE increased promoter activation, both in the presence or absence of LPS and IFNγ. NE alone did not activate the promoter present in C6–2.2 or C6-CREB cells (Fig. 2b).
The above results suggest that the area of the NOS2 promoter extending from bp −187 to −131 is required for NE-suppression of activation by LPS plus IFNγ and also restricts activation by NE. Since this region of the promoter contains several potential transcription binding sites, we prepared an additional construct (pGL3-ΔCREB) lacking the 5′ end 27 bp (Fig. 1). Activation of the NOS2 promoter in these cells was essentially the same as observed for that in C6-κB cells (Fig. 2b), namely strong stimulation by NE alone, and little or no activation due to incubation with LPS and IFNγ. These results suggest that this 27-bp region is a key regulator of NOS2 promoter activation.
Effects of NE on NF-κB activation
Since NOS2 expression in glial cells requires NF-κB activation we tested if NE reduced NF-κB activation by EMSA using the rat NOS2 proximal NF-κB oligonucleotide (Fig. 4a). Nuclear extracts prepared from control (non-stimulated) cells showed low levels of several κB:protein complexes. Following 1 h incubation with LPS and IFNγ, the amounts of two major complexes (bands 3 and 5) were increased. Lower amounts of two other, more slowly migrating complexes were also apparent at 1 h (bands 1 and 2). The levels of several slowly migrating species (arrowheads) remained unchanged by LPS and IFNγ. The presence of NE during the incubation had no effects on its own, nor on the levels of either the stronger (bands 3 and 5) nor weaker (bands 1 and 2) complexes induced by LPS and IFNγ.
We next examined the effects of NE on NF-κB activation after 5 h incubation, a time point closer to, but still preceding the time at which measurements of NOS2 promoter activity were made. At this time, incubation with LPS and IFNγ resulted in the same overall pattern of protein: DNA complexes formed as that observed after 1 h incubation. The major differences noted were a significant increase in the amount of complex 3, while the levels of complex 4 slightly increased and those of complex 5 decreased. As found after 1 h incubation, the presence of NE did not noticeably reduce the levels of either the stronger or weaker complexes. In primary astrocyte cultures, the overall pattern of NF-κB activation due to LPS and IFNγ was simpler than that observed in C6 cells, with only strong induction of complex 3. However, as found for C6 cells, NE did not suppress complex 3 formation (but in fact may lead to a slight increase). Competition experiments (Fig. 4b, and see Stasiolek et al. 2000) demonstrated that the formation of complexes 3 and 4 were abolished by co-incubation with unlabeled sense κB oligonucleotide, but not by a mutated probe. Formation of complexes 1 and 2 were blocked by the sense probe, and to some extent by the mutated probe. Supershift EMSAs suggest that complex 3 consists of a p50 : p65 heterodimer, while complex 4 contains p50 but not p65.
To provide evidence that activation of luciferase expression from the NOS2 promoter was NF-κB dependent, we tested the effects of lactacystin, a highly selective inhibitor of the 26S proteasome which we have shown blocks NF-κB activation in C6 cells (Stasiolek et al. 2000) (Fig. 5). Lactacystin blocked promoter activation by LPS and IFNγ in both C6–2.2 and C6-CREB cells. Surprisingly, promoter activation by NE in C6-κB cells was also blocked by lactacystin, suggesting that NE-dependent activation may also involve protein degradation by the proteasome.
Activation of DNA binding factors by NE
Gel shift analysis of the 58-bp NOS2 promoter region from −187 to −130 was carried out to begin to define transcription factor binding to this area. Extracts from control (non-stimulated) cells demonstrated constitutive binding to this region (Fig. 6a), whose levels were not significantly modified upon incubation with LPS and IFNγ. However, incubation of cells with NE (either in the presence or absence of LPS and IFNγ) led to appearance of one major new complex (indicated by arrow). The presence of a 100-fold molar excess of Oligo3, which contains the consensus CRE and C/EBP sites, but not mutant-oligo3, abolished formation of both constitutive and the NE-induced complexes (Fig. 6b). This results suggests that protein binding to the 58-bp regions occurs primarily to the CRE and C/EBP sites.
Similar results were obtained when a 21-bp oligonucleotide derived from near the 5′ end of this region (bp −178 to −158; see Fig. 1 for sequence) was used as probe (Fig. 7). In control cells, constitutive DNA binding was observed, and the levels of minor complexes were reduced by pre-incubation with antibodies to either CREB or C/EBP (reactive against α, β and δ forms). No differences in the patterns of DNA : protein complex formation were observed due to incubation with LPS plus IFNγ (not shown). Extracts prepared from cells treated with NE (either alone, or in combination with LPS and IFNγ, not shown) caused appearance of one major new band (complex 2), which was reduced by pre-incubation of extracts with an antibody to CREB, but not to C/EBP or Oct-1. NE did not significantly effect the levels of minor complexes.
Effects of NE on IκBα expression
To determine if the inhibitory effects of NE on NOS2 expression could include effects on inhibitory protein IκBα expression, we measured IκBα protein levels after incubation in the presence or absence of NE (Fig. 8). In C6 cells, the IκBα protein was present in non-stimulated cells, and those levels were not significantly altered after 4 h incubation in NE. As previously shown, incubation with LPS and IFNγ decreased IκBα levels at 4 h, due to proteolytic degradation (Stasiolek et al. 1999). However, this decrease was partially blocked by the presence of NE. After 24 h, IκBα levels were slightly increased by NE alone, while levels in LPS plus IFNγ were further reduced by NE. The levels of IκBα in primary astrocytes were greatly increased after 4 h incubation with NE, and as found for C6 cells, NE reduced the decrease due to incubation with LPS. Similar changes were observed at 24 h as at 4 h. These results demonstrate that NE can increase IκBα levels, and minimize or delay the loss due to inflammatory stimuli.
The results presented here confirm and extend our previous findings that in brain glial cells, NE suppresses NOS2 expression at the transcriptional level (Feinstein 1998), and point to the region extending from −187 to −160 of the rat promoter as being critical for two distinct, but likely interrelated, aspects of promoter regulation. First, whereas incubation with LPS and IFNγ activated the 2.2-kb and the 187-bp promoter constructs, these inducers did not activate the 160-bp construct present in ΔCREB cells (nor the 130-bp construct present in C6-κB cells), suggesting that the region from −187 to −160 is an enhancer of cytokine-dependent NOS2 expression. Secondly, whereas activation of the 2.2 kB and 187-bp promoters was reduced by co-incubation with NE, a suppressive effect of NE was not observed when using either the 160-bp or 130-bp promoters (in fact NE increased the activation of these two constructs) supporting the idea that NE suppressive effects are mediated by the − 187 to −160-bp region. Finally, the fact that the 187-bp promoter construct was not activated by NE suggests that transcription factor binding to the −187 to −160-bp region (either constitutively, or induced by NE) also suppresses activation by NE.
Our current findings are consistent with previous results which examined activation of the mouse NOS2 promoter stably expressed in C6 cells (Feinstein 1998). However, in those studies, co-incubation with NE reduced activation of the longer 1588-bp promoter by approximately 50%, in contrast to the roughly 20% reduction observed in the present studies. While the basis for this difference is not clear, we suggest that upstream sequences present in the 2.2 kB rat NOS2 promoter, but absent from the 1588-bp mouse promoter, further contribute to regulation by NE acting in concert with the −187 to −160 region.
The 27-bp promoter region from −187 to −160 is highly conserved between rodent and human promoters, suggesting a role for this area in regulating human NOS2 transcription. Between mouse and rat, there is 24/27 identity, whereas the overall similarity is 80% between rat and human, containing one stretch of 15/16 identities. Although in both mouse and human the CRE site (TGATGTA) differs by one base from the canonical CRE sequence (TGACGTA), the C/EBP site is maintained. This suggests that binding of CREB to this region may also involve interactions with the C/EBP site, as previously shown for the COX-2 promoter (Wadleigh et al. 2000).
Although incubation with LPS and IFNγ did not increase promoter activation in the C6-ΔCREB cells, we did observe some activation, in the C6-κB cells. The region from −160 to −130 bp contains several potential binding sites, including one for the POU transcription factor Brn-2 which activates a variety of brain specific genes (Schreiber et al. 1992), whose binding to this region could reduce the efficacy of NF-κB binding to its cognate element.
Our results indicate that in glial cells, the presence of the proximal NF-κB binding is not sufficient to confer induction by LPS and cytokines, but that additional upstream sequence is needed. Whereas LPS plus IFNγ activated up to 5–6-fold the NOS2 promoter present in C6–2.2 and C6-CREB cells, it failed to activate the promoter in either C6-ΔCREB or C6-κB cells. Similarly, a requirement for the promoter region immediately upstream of the proximal NF-κB site has previously been implicated in several studies. In rat mesangial cells (Eberhardt et al. 1998), the region of the rat promoter extending to bp −277 conferred comparable induction by IL-1β as a larger construct extending to bp −1713, while in RASMCs a mouse promoter construct extending to bp −234 gave 75% of the response to IL-1β as did a promoter extending to bp −1485 (Perrella et al. 1996). A key role for the −187 to − 160 region is suggested by recent mutational studies. Using the mouse promoter for transfections into mouse fibroblast 3T3 cells or macrophage J774.A1 cells, double mutation of the C/EBP2 site (leaving the proximal NF-κB site intact) reduced induction by LPS/IFNγ by over 90% (Dlaska and Weiss 1999). Similarly, mutation of the C/EBP2 region in the mouse promoter reduced the LPS/IFNγ induction by over 50% when transfected into mouse ST-1 epithelial cells (Gupta and Kone 1999). Finally, suppression of LPS/IFNγ dependent NOS2 promoter induction in RAW cells by α-MSH (Gupta et al. 2000) was mediated by reduced binding of C/EBPβ to the C/EBP2 region, but not by inhibition of NF-κB binding. These findings indicate that in certain cell types, transcription factor binding to this portion of the NOS2 promoter is necessary for induction by LPS and cytokines.
In contrast to the above, in rat aortic smooth muscle cells (RASMC), a construct extending to bp −484, provided only slight (1.5-fold induction) activation by LPS or by a cytokine mixture [IL-1β and tumour necrosis factor-α (TNF-α)], compared with the construct extending to −1.7 kB which yielded induction of roughly 13-fold (Zhang et al. 1996). This is similar to results using rat myocytes in which the region upstream of bp −542 of the rat promoter was necessary for strong induction by LPS (Kinugawa et al. 1997). In rat vein smooth muscle cells (VSMCs; Spink et al. 1995), the region upstream of bp − 899 of the mouse promoter was needed for a strong response to LPS, while in studies using RAW cells the region from bp −1029 to −724 of the mouse promoter (Lowenstein et al. 1993) was needed for stimulation by LPS. Hence, these studies point to more distal promoter elements being necessary for induction by LPS and cytokines.
Whereas the above studies implicate promoter regions upstream of the NF-κB site for induction, other studies suggest that a minimal promoter (containing only the proximal NF-κB site) can be activated by LPS or cytokines. In mouse macrophage cells, the p8.11-CAT mouse NOS2 promoter construct, which begins exactly at the 5′ end of the proximal NF-κB site, conferred inducibility by LPS alone in a dose-dependent manner as did the longest promoter construct tested (1588 bp), although the magnitude of induction was about two-fold less (Xie et al. 1994). Similarly, in rat mesangial cells a rat NOS2 promoter (Eberhardt et al. 1998) extending to bp −111, and therefore 4-bp upstream of the proximal NF-κB site, responded similarly (roughly 80%) to stimulation by IL-1β as did a longer construct extending to −526 bp. Thus, while the NF-κB site may be sufficient to achieve low levels of induction, in most cases upstream sequence is needed for high levels of LPS and cytokine dependent NOS2 promoter activation.
Our present results comparing promoter activation by LPS and cytokines vs. cAMP (due to incubation with NE) differ from those obtained using rat mesangial cells. In those cells, NOS2 could be induced by incubation with either IL-1β or cAMP elevating agents alone (Kunz et al. 1994), while co-incubation led to a synergistic increase in total NOS2 expression (Muhl et al. 1994). Molecular analysis revealed that the C/EBP2 site present in the rat NOS2 promoter was indispensable for induction by cAMP, but not needed for induction by IL-1β (Eberhardt et al. 1998). These results are in sharp contrast to those in glial cells, where we find that the presence of the C/EBP2 site (or a nearby area) prevents induction by NE, while this site is necessary to observe strong induction by LPS and cytokines. Despite these opposite effects of cAMP on NOS2 induction in these two cell types, the factors responsible may be related. In mesangial cells, the IL-1β dependent NOS2 expression required NF-κB activation, whereas NOS2 induction due to cAMP was not effected by NF-κB inhibitors, nor did cAMP induce NF-κB activation (Eberhardt et al. 1998). Likewise, incubation of glial cells with NE did not lead to NF-κB activation, as assessed by EMSA (or by staining for nuclear localization of the p65 subunit, not shown). These findings suggest that cAMP-dependent induction of NOS2 in mesangial cells, as well as NE-dependent induction of the minimal NOS2 promoters in glial cells, is not mediated by NF-κB activation. Supershift EMSAs suggested that binding of C/EBPβ, C/EBPδ and/or CREB was responsible for cAMP induced mesangial cell NOS2 expression (Eberhardt et al. 1998), suggesting that activation of these factors could also account for induction of the minimal NOS2 promoter in glial cells, perhaps by binding to the C/EBP3 site that is located downstream of the proximal NF-κB site (see Fig. 1). The fact that in C6 cells, lactacystin, a highly specific inhibitor of the 26S proteasome (Stasiolek et al. 2000), blocked both cytokine as well as NE-dependent NOS2 activation suggests that proteasome dependent degradation is also involved in the NE-dependent activation. It has recently been shown that the 26S proteasome can degrade repressors of cAMP induced transcription such as ICER-IIγ (a product of the CREM gene) and ATF5 (a novel ATF homolog) (Pati et al. 1999), and that lactacystin induced expression of Gadd153 and ATF3, factors which can both bind to and activate C/EBP sites (Zimmermann et al. 2000). Whether these factors also participate in glial NOS2 induction remains to be determined.
Analysis of the factors binding to the C/EBP2 region in mesangial cells revealed both constitutive as well as inducible binding to this region (Eberhardt et al. 1998), as we observed in glial cells. The factors identified in mesangial cells which bound to this region include C/EBPβ, C/EBPδ, as well as CREB. Our studies using antibodies to block complex formation suggest that NE induces binding of CREB, or a CREB-related protein to this region. Since binding of CREB to a C/EBP site present in the COX-2 promoter reduces COX-2 expression (Wadleigh et al. 2000), NE induced CREB binding to the C/EBP in the NOS2 promoter could contribute to its inhibitory effects. Overall, these results indicate that, although the same area of the NOS2 promoter can modulate activation in both glial and mesangial cells, the cell specific factors that are activated by immunostimulation can lead to opposite results.
Our results show that although NE did not reduce NF-κB activation (assessed by EMSA), it increased mRNA and protein levels of the inhibitory IκBα protein. Previous studies have shown that elevation of cAMP, either by βAR agonists (Farmer and Pugin 2000), by peptides including α-MSH (Manna and Aggarwal 1998) or VIP (Delgado and Ganea 2000), or by use of cAMP mimetics (Farmer and Pugin 2000), can increase IκBα levels in immunostimulated cells. However, in contrast to our findings, increases in IκBα levels were accompanied by a reduction of NF-κB activation. Since IκBα can enter the nucleus and associate with DNA-bound NF-κB (Arenzana-Seisdedos et al. 1995), it is possible that in glial cells, inhibitory actions of NE are mediated primarily within the nucleus. However, whether NE increases nuclear levels of IκBα are not yet known. Finally, since our data indicates that suppression by NE requires binding of a factor(s) to the −187 to −160 region, we suggest that potential inhibitory interactions of IκBα with promoter bound NF-κB require complex stabilization by interaction with other factors or cofactors bound to this region. For example, the transcriptional coactivators CBP can bind to CREB as well as NF-κB (Parry and Mackman 1997), while the catalytic subunit of PKA can bind to CREB, IκBα and NF-κB (Zhong et al. 1997).
Our observations confirm and extend previous in vitro results that NE reduces astroglial inflammatory gene expression, as has been demonstrated in numerous other cell types (Galea and Feinstein 1999). Since astrocytes express βARs in vivo (Milner et al. 2000), it is feasible that astroglial inflammatory responses are also regulated by NE in vivo, whose dysregulation could be a factor in human neurological disease. In autoimmune demyelinating disease, a role for NE is suggested by several lines of evidence. In EAE rats, spinal cord NE concentrations are depleted compared with control rats (White et al. 1983), and depletion of central NE levels by lesion of the locus ceruleus (LC) (Konkol et al. 1990; Jovanova-Nesic et al. 1993) influences the course of EAE disease. It has also been shown in rodents that βAR agonists (Chelmicka-Schorr et al. 1989), or selective phosphodiesterase IV inhibitors (Genain et al. 1995; Sommer et al. 1995; Wiegmann et al. 1995; Dinter et al. 2000) suppress EAE symptoms. In humans, βAR expression is higher in peripheral blood monocytes in MS patients than in controls (Karaszewski et al. 1993), again suggesting derangement of the NE : βAR signaling system. Recently, immunocytochemical (De Keyser et al. 1999) and radiolabeling studies (Zeinstra et al. 2000) revealed the presence of β2AR on astrocytes in the white matter of control brains, but not on astrocytes in either normal appearing white matter nor chronic active or inactive plaques in MS brains, providing direct evidence for abnormal βAR expression in MS brain, and suggest that lack of β2ARs on astrocytes in MS white matter could be permissive for inflammatory gene expression in those cells.
Existing evidence suggests that NE may also play a role in modulating inflammatory responses in Alzheimer's disease (AD). An inflammatory component characterized by activation of astrocytes and microglia contributes to neuronal dysfunction and cell death observed in AD (Mann et al. 1980; Floyd 1999), and NOS2 expression has been observed in astrocytes of various brain regions undergoing neurodegeneration (Wallace et al. 1997). An additional characteristic of AD is progressive cell loss of noradrenergic neurons of the locus ceruleus (LC) in the brainstem (Mann et al. 1980; Bondareff et al. 1981). The LC is the major source of noradrenergic innervation of the hippocampus, entorhinal and frontal cortex (Losier and Semba 1993; Loy et al. 1980), the same regions that show the earliest and most pronounced neurodegenerative changes in AD (Westlund et al. 1983). Loss of LC neurons leading to decreased NE content in AD brain could therefore be permissive for Aß-induced NOS2 expression, leading to prolonged NOS2 expression and NO release which could contribute to neuronal dysfunction and mediate cell death.
This work was supported in part by grants from NINDS, NIH (NS-31556; DLF) and the National Mulitiple Sclerosis Society. We thank Anthony Sharp and Patricia Murphy for technical assistance and Dr M. Heneka for critical reading of the manuscript.