Inhibition of lipopolysaccharide-stimulated TNF-α promoter activity by S-adenosylmethionine and 5′-methylthioadenosine

Nary Veal, Chih-Lin Hsieh, Shigang Xiong, Jose M. Mato, Shelly Lu, Hidekazu Tsukamoto


S-adenosylmethionine (SAM) is the principal biological methyl donor and precursor for polyamines. SAM is known to be hepatoprotective in many liver disease models in which TNF-α is implicated. The present study investigated whether and how SAM inhibited LPS-stimulated TNF-α expression in Kupffer cells (hepatic macrophages). SAM downregulated TNF-α expression in LPS-stimulated Kupffer cells at the transcriptional level as suggested by a transfection experiment with a TNF-α promoter-reporter gene. This inhibition was not mediated through decreased NF-κB binding to four putative κB binding elements located within the promoter. The inhibited promoter activity was neither prevented by overexpression of p65 and/or its coactivator p300 nor enhanced by overexpression of coactivator-associated arginine methyltransferase-1, an enzyme that methylates p300 and inhibits a p65-p300 interaction. SAM did not lead to DNA methylation at the most common CpG target sites in the TNF-α promoter. Moreover, 5′-methylthioadenosine (MTA), which is derived from SAM but does not serve as a methyl donor, recapitulated SAM's effect with more potency. These data demonstrate that SAM inhibits TNF-α expression at the level downstream of NF-κB binding and at the level of the promoter activity via mechanisms that do not appear to involve the limited availability of p65 or p300. Furthermore, our study is the first to demonstrate a potent inhibitory effect on NF-κB promoter activity and TNF-α expression by a SAM's metabolite, MTA.

  • Kupffer cells
  • macrophages
  • gene regulation
  • nuclear factor-κB

S-adenosylmethionine (SAM) is the principal biological methyl donor and a precursor for glutathione and polyamines. It is generated after the first reaction in metabolism of methionine catalyzed by methionine adenosyltransferase. In the biosynthesis of polyamines from SAM, 5′-methylthioadenosine (MTA) is generated as a by-product (19). Hepatic deficiency of SAM is a common metabolic abnormality in patients with alcoholic liver disease (ALD) and in experimental models of liver injury caused by cholestasis, ethanol, or other hepatotoxins, such as acetaminophen, galactosamine, and carbon tetrachloride (20, 41). Conversely, a long-term treatment with SAM improves survival or delays liver transplantation in patients with alcoholic liver cirrhosis, and SAM is also hepatoprotective in the experimental models described above (7, 14, 27, 41). Mechanisms of this hepatoprotection are still elusive, but recent studies suggest that they are mediated, at least in part, by inhibition of TNF-α expression (29). TNF-α, a pleiotropic inflammatory cytokine, is implicated in the pathogenesis of liver injury caused by various etiologies (3, 6, 10, 29, 33). Patients with ALD frequently have elevated plasma concentrations of TNF-α and endotoxin, and neutralization of TNF-α or deficiency of TNF-α receptor ameliorates experimental ALD (24, 29, 46). In vitro studies demonstrate that SAM dose dependently decreases LPS-stimulated TNF-α expression in RAW264.7 cells, a murine macrophage cell line, and SAM's preventive effects on experimental liver injury are accompanied by reduced serum levels of TNF-α (10, 43). However, the mechanisms underlying the inhibitory effect of SAM on LPS-stimulated TNF-α expression are yet to be elucidated.

TNF-α gene transcription is largely predicated by binding of NF-κB to its cis regulatory elements within the TNF-α promoter (35). NF-κB encompasses a Rel family of inducible transcriptional activators that are critical in the regulation of genes involved in the inflammatory, immune, and antiapoptotic responses in mammals. The activity of NF-κB is regulated at several levels, the first one being its subcellular localization. In unstimulated cells, NF-κB proteins are sequestered in the cytoplasm through their binding to a family of inhibitory proteins termed inhibitor κBs (IκBs). Cell activation by a multitude of extracellular signals, including LPS, inflammatory cytokines, phorbol esters, UV irradiation, or oxidants, results in the phosphorylation, ubiquitination, and subsequent proteosomal degradation of IκB. This is followed by rapid translocation of liberated NF-κB to the nucleus where it binds to specific κB sites on target gene promoters and stimulates transcription. Four putative κB sites have been identified in the murine TNF-α promoter (5, 13, 18). These sites are termed κB1 to -4 and are located at the nucleotide positions −850, −655, −510, and −210, respectively (5, 13, 18). Each of these sites appear to mediate transcriptional activation by LPS, but κB3 and κB2 have a predominant role. In the cell, NF-κB exists as homo- or heterodimers of the Rel family, which includes p50, p52, p65 (RelA), c-Rel, and RelB. The two most common complexes of NF-κB are the p65/p50 heterodimer and p50/p50 homodimer.

Their binding to DNA constitutes a second level of regulation for NF-κB activity. Binding of p65/p50, but not p50/p50, results in the activation of TNF-α promoter. As for p50/p50, Bohuslav et al. (5) demonstrated that binding of this homodimer to the murine κB3 was responsible, at least in part, for LPS tolerance defined as a silencing of TNF-α transcription after a second challenge of the cells with LPS.

The third level of regulation for the activity of NF-κB takes place through its binding to transcriptional coactivators that function to bridge sequence-specific transcription factors to the basal transcriptional machinery. Cyclic-AMP response element binding protein (CREB)-binding protein (CBP) and its homologue p300 are transcriptional coactivators known to interact with a variety of trans-acting factors including p65. CBP and p300 physically interact with p65, and overexpression of CBP or p300 enhances the transactivation potential of p65 (21, 32, 36). These interactions are influenced by posttranslational modifications of p65 and CBP/p300. For instance, phosphorylation of p65 increases its binding to CBP/p300, whereas acetylation of p65 reduces its ability to bind a κB site (25, 38, 47). More relevant to SAM's potential effects is methylation of CBP/p300 that decreases CBP/p300 binding to transcription factors, such as CREB, and secondary coactivators, such as GRIP1 (11, 45). The enzyme known to be involved in CBP/p300 methylation is coactivator-associated arginine methyltransferase-1 [CARM-1; also called protein arginine methyl transferase (PRMT-4)]. CARM-1 belongs to a family of PRMT methyltransferases (PRMT-1, -2, -3, -5 and -6) and was originally identified as a histone methyltransferase. However, CARM-1 has also been shown to methylate CBP/p300 using SAM as a methyl group donor (11, 45). Therefore, SAM may promote methylation of CBP/p300 by CARM-1 and consequently decrease p300 binding to p65 and the transcriptional activity of p65.

The present study was aimed at determining whether SAM suppresses LPS-stimulated TNF-α expression in Kupffer cells, resident macrophages in the liver, and if so, at examining which aforementioned levels of regulatory mechanisms for NF-κB activity are affected by SAM. More specifically, we examined whether SAM's effect was mediated by changes in NF-κB binding to the putative κB sites in TNF-α promoter or in the promoter activity facilitated by p65 and p300 and modulated by CARM-1. Our results demonstrate that SAM inhibits TNF-α expression at the level of the promoter without modifying NF-κB binding to the κB sites. Expression of CARM-1 does not enhance the inhibitory effect of SAM on TNF-α promoter, and overexpression of p65 and/or p300 does not ameliorate this effect. More importantly, MTA, a metabolite of SAM that cannot serve as a methyl donor, recapitulated SAM's inhibitory effect on TNF-α expression but with higher potency.


Cell lines and reagents.

RAW264.7 and COS-7 cells were obtained from the American Type Culture Collection (Rockville, MD). SAM in the stable form of sulfate-p-toluenesulfonate salt produced by Knoll (Milan, Italy) was purchased from Europharma (Madrid, Spain). All other reagents were of analytical grade, and unless otherwise stated, they were purchased from Sigma (St Louis, MO). [32P]dCTP (3,000 Ci/mmol) was purchased from New England Nuclear (DuPont, Boston, MA).

Kupffer cell isolation.

Kupffer cells were isolated from normal Wistar rats by in situ sequential digestion of the liver with pronase and collagenase and arabinogalactan gradient ultracentrifugation as previously described (28, 40). Kupffer cell viability was tested by the trypan blue exclusion test and always exceeded 97%. The adherence purification method was performed to raise the purity of Kupffer cells >96% as determined by phagocytosis of 1 μm latex beads. Briefly, freshly isolated cells were seeded at 50 × 106 cells/10-cm dish and were incubated for 3 h with DMEM (GIBCO-BRL, Grand Island, NY) containing 10% FBS (GIBCO-BRL), supplemented with 2 mM l-glutamine (Biochrom KG), penicillin (100 U/ml), and streptomycin (100 μg/ml) (Biochrom KG). Nonadherent cells were removed 3 h after seeding, and fresh culture medium with 5% FBS was added and replaced every day until the experiments began. Cells were grown in a humidified 5% CO2-37°C atmosphere and were used 2–3 days after seeding.

Culture experiments.

Unless otherwise specified, Kupffer and RAW cells were pretreated in serum-free DMEM with SAM, methionine, or MTA for 16 h. They were then stimulated with LPS (Escherichia coli 055:B5; Sigma, final concentration: 500 ng/ml) or an equal volume of solvent (water) for either 45 min (assessment of NF-κB/DNA binding) or 4 h (TNF-α mRNA and protein measurements and transient transfection experiments). When needed, cyclohexamide (0.5 μg/ml final concentration) and cycloleucine (10–20 mM final concentrations) were added 1 h before SAM or MTA treatment. Cyclohexamide, a protein biosynthesis inhibitor, was used to check whether SAM's effects required de novo protein synthesis. Cycloleucine is an inhibitor of methionine adenosyltransferase, the enzyme that catalyzes SAM biosynthesis (9) and its effect on SAM levels was tested in RAW, COS-7, and HepG2 cells.

TNF-α immunoassay.

The effect of SAM on LPS-stimulated TNF-α release by cultured Kupffer cells was examined by analyzing TNF-α protein in the culture medium using a commercially available mouse TNF-α immunoassay kit (R&D Systems, Minneapolis, MN).

Ribonucleic acid extraction, Northern blot analysis, and RT-PCR.

To examine the effects of SAM, methionine, or MTA on LPS-stimulated TNF-α mRNA expression, total RNA was isolated from SAM-, methionine-, or MTA-pretreated Kupffer cells 4 h after LPS stimulation. RNA samples were isolated by using TRIzol (Promega, Madison, WI) and after the methodological concept developed by Chomczynski and Sacchi (12). RNA concentration was determined spectrophotometrically, and the integrity was checked by electrophoresis of samples with subsequent ethidium bromide staining. Northern hybridization analysis was performed on total RNA using standard procedures and a [32P]TNF-α cDNA probe as previously described (28, 40). To ensure equal loading of RNA samples and transfer to the membrane, the membrane was rinsed with ethidium bromide and photographed for assessment of 28S and 18S rRNA levels after transfer. Furthermore, the same membrane was rehybridized with a 32P-labeled 18S rRNA probe (39). TNF-α and 18S probes were labeled with [32P]dCTP using a random-primer kit (Primer-It II Kit; Stratagene, La Jolla, CA). Northern blots were exposed to phosphoimager screens, and the radioactive bands were visualized and quantified by using a phosphoimager and ImageQuant software (Molecular Dynamics, Sunnyvale, CA). Results of TNF-α Northern blot analysis were normalized to 18S. To validate overexpression of the gene of interest by transient transfection, total RNA from transfected RAW264.7 cells was also extracted by using TRIzol solution (Invitrogen, Carlsbad, CA). Two micrograms of RNA was reverse-transcribed, and 10% of the first-strand cDNA pool was subjected to PCR amplification by the standard protocol. The following PCR primers were used: 5-AGTCACCGTTGTTTGC-3′ forward and 5′-CCGAGGGAGACGTGTA-3′ reverse for an amplicon of 213 bp for CARM-1; 5′-GGTCCACTCCAATCCAG-3′ forward and 5′-CGTCTCAAGATGTCTCGGAA-3′ reverse for an amplicon of 186 bp for p300; and 5′-ACCCCTTTCACGTTCC-3′ forward and 5′-TCCGGTTTACTCGGCA-3′ reverse for an amplicon of 189 bp for p65. The PCR programs were as follows: 30 cycles at 94°C for 30 s, 58°C for 45 s, and 72°C for 1 min for CARM-1; 31 cycles at 94°C for 30 s, 55°C for 45 s, and 72°C for 1 min for p300; and 22 cycles at 94°C for 30 s, 59°C for 45 s, and 72°C for 60 s for p65 in a DNA thermal cycler (Genius/Techne, Cambridge, UK). The PCR products were separated by electrophoresis on 2% agarose gels and were visualized by ethidium bromide staining with ultraviolet light illumination.

Nuclear protein extraction from Kupffer cells and EMSA.

To examine the effects of SAM on DNA binding by NF-κB, nuclear proteins were extracted from cultured Kupffer cells pretreated with SAM and stimulated with LPS for 45 min, using the method of Schreiber et al. (34). Nuclear extract concentrations were determined by using Bio-Rad Protein Assay dye (Life Science Research, Hercules, CA) based on the Bradford colorimetric assay. Nuclear proteins (5 to 10 μg) were incubated on ice with an oligonucleotide probe labeled with 32P in a reaction mixture [in mM: 20 HEPES, pH 7.6, 100 KCl, 0.2 EDTA, 2 dithiothreitol, 20% glycerol, 200 μg/ml poly(dI-dC)]. For supershift assays, after 20 min of incubation, 1 μl of antibodies against p50 and/or p65 (sc-114X and sc109X, 200 μg/100 μl; Santa Cruz Biotechnology, Santa Cruz, CA) were added to the reaction mixture for an additional 30 min. The reaction mixture (25 μl) was then resolved on a 6% nondenaturing polyacrylamide gel using 0.4 × TBE (in mM: 25 Tris·HCl, 25 boric acid, 0.5 EDTA) as running buffer. Gels were transferred to Whatman 3M paper and dried under vacuum. Protein binding was assessed by using a phosphoimager and the ImageQuant software (Molecular Dynamics, Sunnyvale, CA).

For probe labeling, oligonucleotides described below were labeled with [32P]dCTP using T4-polynucleotide kinase (Boehringer-Mannheim, Mannheim, Germany) and then purified on a NucTrap probe purification column (Stratagene, San Diego, CA). The sequences used for the probes include the consensus κB sequence: 5′-GCAGAGGGGACTTTCCGAGA-3′ (Santa Cruz Biotechnology) and the four putative κB sequences within the murine TNF-α promoter as follows (35): κB1 (−850): 5′-GGGAGGGGAATCCTTGGAAT-3′; κB2 (−655): 5′-GGTCCGTGAATTCCCAGGGC-3′; κB3 (−510): 5′-AACAGGGGGCTTTCCCTCCT-3′; κB4 (−210): 5′-GATCCGGAGGAGATTCCTTGATG-3′. At least three experiments were performed independently for each probe.

Transfection experiments.

To assess the effects of SAM, methionine, and MTA on TNF-α or NF-κB promoter activity, RAW264.7 and COS-7 cells were transiently transfected with a TNF-α promoter or NF-κB-firefly luciferase constructs using Targefect F-2 (Targeting System, San Diego, CA). TNF-α promoter-firefly luciferase construct (designated TNF-α-luc) was created by ligating a 1.4-kb KpnI-HindIII fragment of the murine TNF-α promoter (22) into the pGL3 basic plasmid (Promega). NF-κB promoter-firefly luciferase construct (designated NF-κB luc) was purchased from Stratagene (La Jolla, CA) and contains five repeats of consensus NF-κB binding sequence. Renilla phRL-TK vector was used in our transfection experiments as an internal control for transfection efficiency. Renilla phRL purchased from Promega is a plasmid containing Renilla luciferase gene driven by HSV-TK promoter (abbreviated as Renilla-luc).

For transfection, the cells cultured in six-well plates were treated for 2 h with 1 ml serum-free DMEM transfection mix containing 2 μg of TNF-α-luc or NF-κB-luc, 0.02 μg of Renilla-luc, and 2 μl of F-2 reagent. After 2 h, incubation in humidified 5% CO2-37°C, 1 ml of DMEM with 10% FBS was added to achieve the final FBS concentration of 5% for overnight incubation. Unless otherwise specified, the medium was changed the following day to new DMEM with 10% FBS, and the cells were incubated for an additional 8 h. The medium was then changed to serum-free DMEM with or without SAM or MTA. After 16-h pretreatment with SAM or MTA, cells were stimulated with LPS (500 ng/ml) or an equal volume of solvent (water) for 4 h. Cell lysate was collected at 24 h for luciferase assay by Dual-Luciferase Reporter Assay System (Promega) as recommended by the manufacturer. Briefly, cells were washed with PBS and lysed with the passive lysis buffer. Aliquots of the cell lysates were sequentially assessed for firefly and Renilla luciferase activities using a MicroBeta TriLux (Wallac) luminometer. Levels of reporter gene expression (TNF-α-luc or NF-κB-luc) were normalized by Renilla-luc activity. Four experiments were performed independently, and each experiment was performed in triplicate. As specified in cotransfection experiments with a p65 expression vector, SAM was added to the cells 2 h after the beginning of transfection. This time-point was chosen to expose the cells to SAM before p65-driven TNF-α or NF-κB promoter activity reached a plateau (∼16 h; unpublished data). The following day, cells were stimulated with LPS (500 ng/ml) or an equal volume of solvent (water) for 4 h. Cell lysates were collected at 24 h for luciferase assay using Dual-Luciferase Reporter Assay System (Promega) as previously described.

p65, p300, or CARM-1 cotransfection experiments.

To verify whether overexpression of p65, p300, and CARM-1 would circumvent or enhance the inhibitory effect of SAM on NF-κB promoter activity, RAW264.7 and COS-7 cells were transiently cotransfected with NF-κB-luc plus a p65, p300, or CARM-1 expression vector. We hypothesized that overexpression of p65 and its coactivator p300 would circumvent SAM's effect if SAM interferes p65-p300 interaction (21, 32, 36). On the other hand, CARM-1 expression would enhance SAM's effect if SAM mediates its effects via CARM-1-catalyzed methylation of p65 or p300. A CARM-1 overexpression experiment was performed in RAW264.7 cells cultured in six-well plates using the aforementioned 48-h transfection protocol with the following modifications: transfection mix consisted of 1 ml serum-free DMEM containing 1 μg of NF-κB-luc, 0.02 μg of Renilla-luc, 1 μg of CARM-1 expression plasmid, and 2 μl of F-2 reagent. P300 and p65 overexpression experiments were performed in RAW264.7 and COS-7 cells transfected with 1 ml serum-free DMEM containing 0.5 μg NF-κB-luc, 0.3 μg p65, 2.7 μg p300 plasmids, 0.035 μg Renilla-luc, and 4 μl F-2 reagent. The amounts of plasmids were adjusted to achieve optimal transfection efficiency and a maximal synergistic effect between p65 and p300. In cotransfection experiments without CARM-1, p65, or p300, an equal amount of empty expression vector (pSG5) was used. The expression plasmid for p65 was generously provided by Dr. Ebi Zandi (University of Southern California), and the expression plasmids for p300 and CARM-1 were kind gifts from Dr. Michael Stallcup (University of Southern California).

TNF-α promoter methylation analysis.

To evaluate whether SAM affects the methylation status of the 1.4-kb murine TNF-α promoter, plasmid bearing this promoter was used for transient transfection experiments. After transfection, plasmid DNA was harvested by the Hirt method (23) and assessed by Southern blot analysis after being digested with HpaII, a methylation-sensitive restriction enzyme, as described previously (30).

SAM measurement.

SAM levels were measured in RAW264.7, COS-7, and HepG2 cells using a method previously described (36) with slight modifications. Cells were scrapped off, counted, and homogenized in PBS, and an aliquot was saved for protein assay. The rest was treated with 100 μl of 1 M perchloric acid (PCA) on ice for 5 min and centrifuged at 1,000 g for 15 min at 4°C. The aqueous layer was quantitatively removed, neutralized with 3 M KOH, and centrifuged at 3,000 g for 10 min at 4°C. SAM levels were determined in the neutralized PCA extracts by HPLC (LC-10ATVP pump, SCL-10AVP system control; Shimadzu) with a SPD-10AVP UV detector and a SIL-10ADVP autosampler (Shimadzu) using a Partisil SCX 10-μm column (25 × 0.44 cm ID; Whatman, Cleveland, OH). SAM was eluted isocratically at 1 ml/min with 0.19 M NH4H2PO4 adjusted to pH 2.6 with 2 M H3PO4. SAM levels were calculated by using a standard curve of SAM prepared at the same time as the samples.

Statistical analysis.

Numerical data were expressed as means ± SD, and comparison between treated and control groups was performed by Student's t-test, if the comparison was between two groups, and by ANOVA followed by Fisher's test, if the comparison was between more than two groups. For changes in mRNA, ratios of TNF-α to 18S RNA densitometric values were compared. Significance was defined by P < 0.05.


SAM dose dependently inhibits LPS-induced TNF-α expression in cultured rat Kupffer cells.

Stimulation of Kupffer cells with LPS (500 ng/ml) markedly increased release of TNF-α protein into the medium and its mRNA levels in rat cultured Kupffer cells as expected. Pretreatment with SAM for 16 h dose dependently inhibited TNF-α expression at both protein and mRNA levels (Fig. 1, A and B). The SAM concentration required to achieve significant inhibition started at 250 μM, and the effect was more conspicuous between 750 μM and 1 mM. These concentrations of SAM did not affect cell viability as determined by Syntox Green and Hoesch staining (data not shown). SAM was previously shown by Watson et al. (43) to dose dependently decrease TNF-α mRNA levels in RAW264.7 cells stimulated with LPS for 8 h. We extended this study to demonstrate the similar effect of SAM in primary cultures of Kupffer cells, the most abundant macrophages in the body. Furthermore, we chose to assess the effect after 4 but not 8 h of LPS treatment, because induction of TNF-α transcription begins to dissipate after 4 h (Ref. 14 and our own data not shown). In summary, we confirmed the inhibitory effect of SAM on TNF-α expression in Kupffer cells at the peak of LPS-mediated induction. We also performed the identical experiment using RAW264.7 cells that produced similar results (data not shown).

Fig. 1.

S-adenosylmethionine (SAM) dose dependently decreases release of TNF-α protein (A) and TNF-α mRNA levels (B) in LPS-stimulated Kupffer cells. Three-day cultured Kupffer cells isolated from normal Wistar rats were pretreated 16 h with SAM at indicated doses in serum-free DMEM, followed by a treatment with LPS (500 ng/ml) for 4 h. Release of TNF-α in culture medium was measured by using a commercially available TNF-α immunoassay kit and total RNA extracted for Northern blot analysis. A: TNF-α production data are presented as means ± SD as expressed as %control (without SAM treatment). *P < 0.05 compared with the control (n = 4). B: representative autoradiograms of TNF-α (top) and 18S rRNA (bottom) are shown. The bar graph depicts SAM-induced changes in relative TNF-α mRNA expression as computed by standardization of densitometric data for TNF-α mRNA with those of 18S rRNA and expresses the data as the percentage of the control. *P < 0.05 compared with the control (n = 7).

SAM requires 8-h pretreatment to render its effect on LPS-induced TNF-α expression.

We then examined the effects of the duration of a pretreatment with SAM (750 μM) on LPS-induced TNF-α mRNA expression. In this experiment, the SAM pretreatment time varied from 4 to 24 h, whereas the duration of LPS stimulation was maintained constant (4 h). As shown in Fig. 2, SAM's inhibitory effect required a pretreatment for 8 h or longer to reach the maximal inhibition (60–70%).

Fig. 2.

SAM requires 8-h or longer durations of treatment to inhibit TNF-α mRNA expression in LPS-stimulated Kupffer cells. The experiment was carried out as described for Fig. 1. Three-day cultured Kupffer cells isolated from normal Wistar rats were pretreated overnight with SAM (750 μM) in serum-free DMEM for different times. LPS (500 ng/ml) or equal volume of solvent (water) was then added to the cells for an additional 4 h. Total RNA was then extracted for Northern blot analysis as described in materials and methods. Top, autoradiographs of TNF-α experiment. Comparable results were obtained in additional experiments. *P < 0.05 compared with the control (n = 5).

SAM's inhibitory effect is dependent on de novo protein synthesis.

Such prolonged pretreatment required for SAM's inhibitory effect suggests that the effect might be dependent on de novo protein synthesis. To test this hypothesis, the effect of cyclohexamide (0.5 μg/ml), a protein synthesis inhibitor, on SAM-mediated inhibition of TNF-α mRNA expression was examined. Cyclohexamide treatment enhanced TNF-α mRNA levels in LPS-stimulated Kupffer cells as previously reported (48) (Fig. 3, lane 1 vs. lane 3). This treatment abrogated the inhibition caused by SAM (Fig. 3), suggesting that SAM's effect is indeed dependent on de novo protein synthesis.

Fig. 3.

Cyclohexamide (Cyclo) blocks the inhibitory effect of SAM on TNF-α mRNA expression in LPS-stimulated Kupffer cells. The experiment was carried out exactly as described for Fig. 2 except Kupffer cells were pretreated with cyclohexamide (0.5 μg/ml) or its vehicle in serum-free DMEM 1 h before 16-h incubation without or with SAM (750 μM) and subsequent LPS stimulation (500 ng/ml) for 4 h. Total RNA was then extracted for Northern blot analysis. A representative set of autoradiograms from 3 separate experiments is shown for TNF-α (top) and 18S rRNA (bottom).

SAM does not decrease NF-κB binding to the four putative κB sites located in TNF-α promoter.

LPS stimulation of RAW264.7 or Kupffer cells is known to induce TNF-α mRNA expression through a rapid and transient NF-κB binding to its κB sites within TNF-α promoter. The positive transcription factor is p65/p50, and p50/p50 is considered the inactive or negative counterpart. To assess whether SAM's inhibitory effect on LPS-stimulated TNF-α mRNA expression was due to a decrease in p65/p50 binding to its κB sites, nuclear proteins were extracted from Kupffer cells pretreated with SAM (750 μM, 16 h) and stimulated with LPS (500 ng/ml) for 45 min. EMSA was performed to assess NF-κB binding to four putative κB sequences located within the murine TNF-α promoter (Fig. 4A) as well as to the consensus κB sequence. As shown in Fig. 4B, LPS stimulation increased the intensity of a band in EMSA gels for all probes tested, and this corresponded to p65/p50 binding. Another slightly faster-moving band with an increased intensity was observed with κB1, -3, and -4 and the consensus κB probes, and this corresponded to p50/p50 binding. The specificity of this binding was determined by supershift assay using antibodies against each of NF-κB subunits (Fig. 4C). As reported by others (5, 18), we also observed that NF-κB binding to κB3 and -1 was stronger than that to κB2 and -4 sites. However, we did not observe any appreciable inhibitory effects of SAM (750 μM, 16 h) on LPS-induced p65/p50 binding to all 5 κB probes examined. Furthermore, the binding of p50/p50 that may serve as a negative regulator, particularly for the κB3 site, was not affected by SAM treatment. These results were obtained with nuclear extracts collected 45 min after LPS stimulation in accordance with the known kinetic of LPS-induced p65/p50 binding. However, p50/p50 binding was reported to increase progressively with time (5, 13). To rule out a possible effect of SAM on p50/p50 binding at a latter time point, we performed additional EMSA with nuclear extracts from cells stimulated with LPS for 4 h. However, results were basically the same as those obtained from the 45-min samples, and no change in LPS-stimulated p50/p50 binding was observed in SAM-treated cells (data not shown). In conclusion, SAM does not change the binding of p65/p50 or p50/p50 to the κB sites, regardless of the length of LPS treatment.

Fig. 4.

SAM does not affect binding of NF-κB to 4 TNF-α-specific κB binding sequences (κB-1 to -4) and a consensus κB sequence. Cultured Kupffer cells were pretreated 16 h with SAM (750 μM) or its vehicle in serum-free DMEM, followed by the addition of LPS (500 ng/ml) or its vehicle and incubation for an additional 45 min. A: nuclear proteins were extracted and incubated with oligonucleotide probes: putative 4-κB sites within the murine TNF-α promoter (κB-1 to -4) as schematically depicted according to publications by Drouet et al. (18) and Bohuslav et al. (5) and a consensus κB sequence. The reaction mixture was then resolved on an EMSA gel. B: representative data are shown and comparable results were obtained from at least 3–4 experiments and for κB3 and κB consensus probes; at least 10 separate experiments were performed. Note no differences caused by SAM for binding of p65/p50 or p/50/p50 to any of these probes. C: EMSA supershift experiment was performed by using polyclonal antibodies recognizing NF-κB proteins (p50 or p65).

SAM downregulates LPS-stimulated TNF-α promoter activity.

We then examined whether SAM inhibited LPS-stimulated TNF-α promoter activity by performing transient transfection in RAW264.7 cells with a TNF-α promoter-luciferase construct (TNF-α-luc). Because TNF-α promoter activity is largely dependent on κB elements, we also assessed the effects of SAM on NF-κB promoter using a reporter gene with 5 × κB sequences (NF-κB-luc). All results were standardized by transfection efficiency as determined by cotransfection of a Renilla luciferase plasmid. LPS stimulation increased the TNF-α promoter activity approximately fourfold, and a similar induction was also evident for NF-κB promoter activity (Fig. 5). SAM pretreatment (750 μM, 16 h) attenuated the increases in LPS-stimulated TNF-α and NF-κB promoter activities by ∼35–45%.

Fig. 5.

SAM decreases TNF-α and NF-κB promoters in LPS-stimulated RAW264.7 cells. RAW264.7 cells were transiently transfected with 2 μg of TNF-α promoter-luciferase or consensus NF-κB promoter-luciferase plasmid and 0.02 μg of Renilla phRL-TK, using Targefect F-2 reagent as described in materials and methods. SAM pretreatment (750 μM) started 24 h after the addition of the vectors for 16 h. The following day, LPS (500 ng/ml) or vehicle (water) was added for 4 h. Cell lysates were collected for luciferase assay as described in materials and methods. Levels of TNF-α luc or NF-κB luc activities were normalized by transfection efficiency as determined by Renilla luciferase activity. Note SAM equally inhibits both TNF-α and NF-κB promoter activities induced by LPS. *P < 0.05 compared with the control (n = 8).

SAM does not induce methylation at the HpaII sites in the TNF-α promoter.

Because SAM serves as a methyl donor and methylation has a silencing effect on promoters, we examined whether SAM increased methylation of the TNF-α promoter. This analysis was performed by digestion by HpaII of TNF-α-luc plasmid DNA harvested from transfected RAW264.7 cells that have been pretreated or unpretreated with SAM (750 μM, 16 h) followed by 4-h LPS stimulation. The 1.4-kb TNF-α promoter used in our transfection experiment contained at least 22 CpG sites and four HpaII sites. HpaII would only digest the unmethylated sites, and methylation at these sites would give rise to increased-size HpaII fragments. Southern blot analysis of HpaII-digested DNA did not reveal any change of methylation at the HpaII sites under SAM treatment (data not shown).

CARM-1 overexpression does not enhance SAM's inhibitory effect on κB promoter.

SAM serves as a methyl donor for p300 methylation by CARM-1 and methylated p300 inhibits a p65-p300 interaction and their association with the transcriptional machinery (11, 45). If this mechanism underlies SAM-induced inhibition of κB promoter activity, we would expect to see increased inhibition by concomitant CARM-1 overexpression. This possibility was tested by cotransfecting RAW264.7 cells with NF-κB-luc and a CARM-1 expression plasmid, followed by treatment with SAM and LPS. Increased expression of CARM-1 by the plasmid was confirmed by RT-PCR of RNA samples extracted from transfected RAW cells as shown in Fig. 7D, left. Indeed, CARM-1 overexpression alone decreased the promoter activity by 22% (Fig. 6, bar 1 vs. bar 3), suggesting a possibility that CARM-1 might have exerted this inhibitory effect by using endogenous SAM. On the other hand, SAM's inhibitory effect was not enhanced by CARM-1 overexpression, suggesting that SAM's effect may not be mediated through CARM-1-dependent methylation of p300. However, it is possible that endogenous CARM-1 activity was sufficient to achieve maximal methylation of p300.

Fig. 6.

Overexpression of coactivator-associated arginine methyltransferase (CARM)-1 does not enhance the inhibitory effect of SAM on NF-κB promoter activity in LPS-stimulated RAW264.7 cells. RAW264.7 cells were transiently transfected as described for Fig. 5, except that 1 μg of CARM-1 expression or empty plasmid was cotranfected with NF-κB-luc and Renilla phRL-TK. Note CARM-1 expression slightly inhibits LPS-induced NF-κB promoter activity but did not enhance SAM's inhibitory effect. *P < 0.05.

Overexpression of p65 and its coactivator p300 does not circumvent SAM's inhibitory effect.

To further test whether SAM's suppressive effect on κB-promoter activity involves inhibitory modifications of p65 or p300 or decreased p65-p300 interaction, we then examined whether overexpression of p300, p65, or both would circumvent the inhibitory effect of SAM. In RAW264.7 cells, p65 overexpression drastically induced NF-κB-luc promoter activity and LPS did not achieve a further increase in this activity, probably because overexpressed p65 maximally drove the promoter (Fig. 7A). Unfortunately, our attempt to synergize p65-driven κB-promoter activity by overexpression of p300 failed in RAW264.7 cells (Fig. 7A), making these cells an inappropriate model to test the effect of p300 on the effect of SAM. Such a synergistic effect of p300 was reported for COS-7 in which the evidence for a physical interaction between p300 and p65 was also demonstrated (21, 32, 36). Thus we performed the similar experiment in COS-7 cells. As shown in Fig. 7B, although p300 overexpression alone did not increase the promoter activity; it synergistically enhanced p65-driven promoter activity. These results are consistent with the findings by Gerritsen et al. (21). The differential effects observed in the two cell types may be attributable to several potential reasons. It is possible that p300 was not sufficiently expressed in RAW264.7 cells. However, our RT-PCR analysis confirms a conspicuous increase in p300 mRNA level in the transfected RAW264.7 cells (Fig. 7D). Furthermore, this analysis also shows the relatively low level of endogenous p300 in RAW cells. At any rate, we have established a model with COS-7 in which a synergism between p65 and p300 is inducible. With the use of this model, the effects of p65 and p300 overexpression on SAM's inhibitory effect on the κB-promoter activity was tested. We did not use LPS in COS-7 cells, because these cells do not respond to LPS stimulation in contrast to macrophages. Figure 7C depicts that SAM (750 μM for 16 h) decreased κB-promoter activity to the similar extent not only in the cells transfected with an empty vector but also in those cotransfected with p65 and p300 plasmids, suggesting that overexpression of p65 and p300 does not overcome SAM's inhibitory effect. Similar results were also obtained with cells transfected with p65 only (data not shown). However, these results do not rule out a possibility of SAM-mediated methylation of p300 as a cause of suppressed κB- promoter activity, because methylated p300 cannot interact with p65, even if both proteins are overexpressed. Lastly, we also tested the effect of SAM on p65-driven NF-κB promoter activity in RAW cells. As shown in Fig. 7E, SAM exerted a similar inhibitory effect on the promoter to that seen under LPS stimulation.

Fig. 7.

A: p300 Overexpression failed to enhance p65-driven NF-κB promoter activity in RAW264.7 cells. RAW264.7 cells were transiently cotranfected as previously described with 0.5 μg NF-κB luc, 0.3 μg p65 plasmid, 2.7 μg p300 plasmid, and 0.035 μg Renilla phRL-TK. For appropriate controls, equal amounts of an empty expression vector (pSG5) were used. Note p65 expression was ∼4 times more effective than LPS in inducing the NF-κB promoter activity. This induction was not further enhanced by either LPS or p300 overexpression (n = 3). B: p65 and p300 have a synergistic effect on NF-κB promoter in COS-7 cells. The identical transfection experiment was carried out in COS-7 cells except LPS stimulation. Note p300 overexpression alone does not increase the basal promoter activity but synergistically enhance p65-driven promoter activity (*P < 0.05, n = 3). C: p65 and p300 overexpression does not prevent SAM-induced inhibition of NF-κB promoter in COS-7 cells. The transfection protocol was identical to that described above, except that SAM (750 μM) or its vehicle was added 2 h after the addition of the vectors for a 22-h incubation. For a comparison of the magnitude of SAM-induced inhibition, the data are expressed as the percentage of respective control. Note overexpression of p300 and p65 failed to abrogate the SAM's inhibitory effect. *P < 0.05 (n = 3) D: CARM-1, p300, and p65 expression is increased by transfection. To ascertain whether transfection of expression plasmids increased the expression of CARM-1, p300, and p65, RT-PCR was performed on RNA samples extracted from RAW264.7 cells transfected with respective plasmids. E: SAM inhibits p65-driven NF-κB promoter activity in RAW264.7 cells. RAW264.7 cells were transfected with the NF-κB reporter gene and the p65 expression vector as already described and treated with SAM from 2 h after the addition of the plasmids. Note SAM's inhibitory effect is reproduced in RAW264.7 cells. NS, not significant.

MTA is more potent than SAM and methionine in suppressing κB promoter activity.

We then examined whether SAM's effect could be recapitulated by MTA, a metabolite of SAM that does not serve as a methyl donor. As shown in Fig. 8A, regardless of whether COS-7 cells were cotransfected with p65 and/or p300, MTA (500 μM) decreased NF-κB-promoter activity, and this inhibition was even more potent than that achieved by a higher concentration of SAM (750 μM). Because MTA can be converted to SAM via methionine, we considered the use of cycloleucine, an inhibitor of a key enzyme that catalyzes this conversion. We first tested whether cycloleucine effectively reduced the endogenous SAM level in COS-7 cells. To our surprise, the treatment of COS-7 cells with 20 mM cycloleucine did not significantly affect the SAM level (1.59 ± 0.29 vs. 1.26 ± 0.18 nmol/mg protein) even after a 48 h incubation, whereas the identical treatment for 24 h caused a 60% reduction in HepG2 cells (1.21 ± 0.49 vs. 0.50 ± 0.03 nmol/mg protein). In RAW264.7 cells, the cycloleucine treatment only mildly reduced the SAM level by 35% (0.23 ± 0.07 vs. 0.15 ± 0.01 nmol/mg protein). Thus these results demonstrated that cycloleucine has limited inhibitory effects on COS-7 and RAW cells, and this inhibitor cannot be reliably used in these cell types. We then compared the effects of MTA, SAM, and methionine (a precursor of SAM) on LPS-induced NF-κB-promoter activity in RAW264.7 cells. Whereas methionine (500 μM) did not significantly inhibit the promoter activity, both SAM (750 μM) and MTA (500 μM) decreased it by 50–60%, and MTA tended to suppress more than SAM, despite its lower concentration (Fig. 8B). Finally, we tested the effects of MTA on LPS-induced TNF-α mRNA levels in primary cultures of rat Kupffer cells. As shown in Fig. 8C, MTA maximally reduced the TNF-α mRNA levels at the concentration as low as 250 μM, whereas it required SAM at 500 μM or higher concentrations to achieve the similar maximal effect (data on only 500 μM shown for SAM).

Fig. 8.

A: 5′-deoxy-5′-methylthioadenosine (MTA) and SAM decrease p65-driven NF-κB promoter activity in COS-7 cells, regardless of whether they are cotransfected with p300 and p65. COS-7 cells were transiently transfected as described in materials and methods. SAM (750 μM) or MTA (500 μM) was added 2 h after the addition of the vectors. Experiments were terminated at 24 h. The data are expressed as the percentage of respective control to compare the magnitude of inhibition by SAM or MTA. Note both SAM and MTA inhibit the promoter activity, regardless of whether p300, P65, or both are overexpressed. Also note that MTA's effect is more potent than SAM's despite the lower concentration used (*P < 0.05, n = 3) B: MTA decreases LPS-stimulated NF-κB promoter activity more effectively than SAM and methionine (Met) in RAW264.7 cells. RAW264.7 cells were transfected with NF-κB-luc and Renilla phRL-TK and stimulated with LPS as described in materials and methods. Note MTA (500 μM) inhibits the promoter more than SAM (750 μM), whereas methionine fails to render a significant effect. (*P < 0.05, n = 3) C: MTA reduces LPS-induced TNF-α mRNA expression more potently than SAM in cultured Kupffer cells. Kupffer cells were pretreated for 16 h with SAM or MTA at the indicated concentrations and stimulated with LPS (500 ng/ml) for 4 h. Total RNA was extracted for Northern blot analysis. Note MTA at 250 μM is as effective as SAM at 500 μM in reducing TNF-α mRNA levels, demonstrating the higher potency of MTA (n = 3).


The inhibitory effect of SAM on LPS-stimulated TNF-α expression was previously demonstrated in RAW264.7 cells by Watson, et al. (43). These authors also demonstrated that SAM's inhibitory effect was not due to a change in TNF-α mRNA stability (43). The present study extends this finding to primary cultures of rat Kupffer cells and further demonstrates that SAM's inhibitory effect is mediated by suppressing TNF-α or NF-κB promoter activity but not by changes in NF-κB binding to the putative four κB elements within the TNF-α promoter. Taken together, we conclude that SAM's effect is unlikely to be mediated via either upstream LPS signaling or posttranscriptional mechanisms but largely reflects a transcriptional repression.

Because SAM is the principal methyl group donor, we investigated whether SAM's inhibitory effect on LPS-induced TNF-α transcription involved a methylation process. Both cis-elements (i.e., the promoter) and trans-elements (proteins recruited on the transcription site) may be methylated to result in decreased transcription. Indeed, methylation of TNF-α promoter causes a dramatic loss of its activity (30). However, our analysis using HpaII, a methylation-sensitive endonuclease, revealed no change in methylation at the HpaII sites within the promoter after SAM treatment. Concerning methylation of trans-elements, p65 and p300 are potential targets. The p65 is a crucial component recruited onto the TNF-α transcription site via p300 as a coactivator (21). CARM-1 methylates p300 using SAM as a methyl donor, and this methylation decreases the ability of CBP/p300 to bind transcription factors such as CREB and secondary coactivators such as GRIP1 (11, 45). Similar regulation by CARM-1 and SAM may take place for p300 and p65. However, overexpression of the p300 methylase CARM-1 did not enhance the inhibitory effect of SAM on NF-κB promoter activity. Conversely, overexpression of p300 did not overcome SAM's inhibitory effect. Lastly, whereas concomitant overexpression of p300 and p65 led to an expected synergy for induction of NF-κB promoter activity in COS-7 cells, SAM still exerted a similar magnitude of suppression on the promoter. However, it is possible that p300 is maximally methylated by endogenous CARM-1 and SAM so that transduction of CARM-1 or the addition of exogenous SAM may not affect the extent of inhibition. Furthermore, overexpression of p65 and/or p300 may not correct the defect, because they may be quickly methylated and may not be able to interact. We also demonstrated that MTA reproduced the SAM's effect. However, we could not use cycloleucine to determine whether MTA exerted this inhibitory effect without its conversion to SAM, because this inhibitor was found to minimally suppress the SAM levels in the cell types we have studied. Therefore, the present study could not allow us to directly investigate the potential contribution of p300 or p65 methylation to the inhibitory effects on NF-κB promoter rendered by SAM or MTA. To directly address this question, the methylation status of p300 or p65 and the interaction between the two need to be addressed in SAM-treated cells. Nevertheless, our data on MTA are intriguing in that it appeared to have a more potent inhibitory effect on NF-κB promoter activity and TNF-α mRNA expression than SAM. MTA is a product of SAM metabolism in the polyamine pathway (19). Exogenous SAM can also undergo nonenzymatic hydrolysis in vivo into MTA and homoserine (37, 44). It has been suggested that the beneficial effects of SAM on liver damage could be attributed, in part, to its conversion to MTA (31, 37). We have also reported that MTA mimics SAM's effect on gene expression in cultured rat hepatocytes (26) and apoptosis in liver cells (1). In contrast to SAM, MTA does not contribute to GSH synthesis, is not a methyl donor, and inhibits methyltransferases (15). A lower concentration of MTA was necessary to achieve a comparable inhibitory effect of SAM. This could be attributed, in part, to the differential intracellular availability of both compounds, SAM being a charged molecule and MTA, a noncharged compound. Alternatively, SAM's effects may depend on its conversion to MTA, either spontaneously or via enzymatic catalysis. It is also possible that SAM and MAT have distinct methylation dependent and independent mechanisms of action. Metabolic pathways discussed above for generation of SAM and MTA are depicted in Fig. 9.

Fig. 9.

A schematic diagram depicting metabolic pathways for SAM and MTA. SAM is synthesized from methionine and ATP by methionine adenosyltransferase. Whereas SAM generates S-adenosylhomocysteine (SAH) via a transfer of its methyl group catalyzed by SAM methyltransferase, SAM, after decarboxylation, also acts as a propylamino donor for the biosynthesis of polyamines. Polyamine (spermine and spermidine) synthesis from SAM yields MTA as a by-product. MTA can be recycled back to SAM via methionine synthesis.

Because SAM did not affect NF-κB binding to its elements, it is likely that SAM or its metabolites mediates its effect through the recruitment of components of TNF-α transcription machinery. To this end, our finding that SAM's inhibition was dependent of de novo protein synthesis, indicates that the “component(s)” may need to be synthesized to mediate SAM's effect on TNF-α transcription. SAM has recently been shown to induce IL-10 expression in RAW264.7 cells by McClain et al. (29). Because this Th2 cytokine is known to suppress macrophage functions including expression of proinflammatory cytokines such as TNF-α and IL-1 (2, 16, 17, 42), IL-10 may be a protein that needs to be synthesized to have SAM's effect. The mechanism of IL-10-mediated suppression of cytokines is diverse and a topic of controversy. It appears to have both transcriptional and posttranscriptional effects (2, 16, 17, 42). In RAW264.7 cells simulated with LPS, suppression of NF-κB promoter activity by IL-10 is associated with inhibition of IκBα degradation and NF-κB binding, (17). However, we did not observe inhibition of NF-κB binding in SAM-treated Kuppfer and RAW264.7 cells exposed to LPS, presenting a dissociation of our data from IL-10's known effect on NF-κB. IL-10's posttranscriptional effects include destabilization of TNF-α mRNA (4, 8) but Watson et al. (43) previously demonstrated that SAM did not affect the TNF-α mRNA degradation, depicting again a discrepancy between the effect of SAM and IL-10 on this level of regulation. Therefore, IL-10 that may be inducible by SAM (29) cannot fully explain the mode of SAM-mediated inhibition of TNF-α expression disclosed by the present study. Further studies are obviously required to identify a de novo synthesized protein(s) that mediate(s) the SAM inhibitory effect on TNF-α promoter activity.


This study was supported by National Institute on Alcohol Abuse and Alcoholism Grants R37-AA-006603 (to H. Tsukamoto and N. Veal), RO1-AA-12677 (to S. Lu, H. Tsukamoto, and N. Veal), RO1-AA-013847 (to S. Lu and H. Tsukamoto), P50-AA-11999 (to Research Center for Alcoholic Liver and Pancreatic Diseases and H. Tsukamoto), R24-AA12885 (to Non-Parenchymal Liver Cell Core and H. Tsukamoto); National Institute of Diabetes and Digestive and Kidney Diseases Grant P30-DK-48522 (to Research Center for Liver Diseases USC, and S. Lu); Medical Research Service of Department of Veterans Affairs Grant (to H. Tsukamoto); a La Fondation pour la Recherche Medicale Grant (to N. Veal); and a Philippe Fondation Grant (to N. Veal).


Present address of N. Veal: Laboratory HIFIH, Faculty of Medicine of Angers, France.


We wish to thank Drs. Michael Stallcup and Ebi Zandi for providing expression plasmids for CARM-1, p300, and p65. We are also grateful to Jiaohong Wang, Hongyun She, Sharon Yavrom, and Saswati Hazra for technical support and helpful discussions on the study.


  • * S. Lu and H. Tsukamoto share senior authorship.

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract