Transforming Growth Factor-β1 Inhibits Non-pathogenic Gramnegative Bacteria-induced NF-κB Recruitment to the Interleukin-6 Gene Promoter in Intestinal Epithelial Cells through Modulation of Histone Acetylation*

We have shown that non-pathogenic enteric Gramnegative Bacteroides vulgatus induces RelA phosphorylation, NF-κB activation, and proinflammatory gene expression in primary and intestinal epithelial cell (IEC) lines. We now demonstrate the transient induction of nuclear phospho-RelA (day 3) followed by persistent activation of phospho-Smad2 (days 3 and 7) in IEC from mucosal tissue sections of B. vulgatus-monoassociated rats, indicating that both NF-κB and transforming growth factor-β1 (TGF-β1) signaling are induced in vivo following bacterial colonization. Interestingly, TGF-β1 inhibited B. vulgatus- and lipopolysaccharide (LPS)-induced NF-κB transcriptional activity as well as interleukin-6 (IL-6) mRNA accumulation and protein secretion in IEC. The inhibitory effect of TGF-β1 is mediated independently of B. vulgatus/LPS-induced IκBα, Akt, and RelA phosphorylation as well as NF-κB DNA binding activity. Moreover, the specific histone deacetylase inhibitor trichostatin A blocked B. vulgatus/LPS-induced histone acetylation/phosphorylation (Lys-9/Ser-10) and reversed TGF-β1-mediated inhibition of IL-6 gene expression. Chromatin immunoprecipitation analysis revealed that B. vulgatus/LPS-induced RelA recruitment to the IL-6 promoter is inhibited by TGF-β1 treatment. Adenoviral delivery of Smad7 and dominant negative Smad3 (SmadΔ3) reversed the TGF-β1-mediated inhibition of NF-κB transcriptional activity and NF-κB recruitment to the IL-6 promoter. In addition, TGF-β1 and Ad5Smad3/4 prevent B. vulgatus/LPS-induced CBP/p300 and p65 nuclear co-association. We concluded that the TGF-β1/Smad signaling pathway helps maintain normal intestinal homeostasis to commensal luminal enteric bacteria by regulating NF-κB signaling in IEC through altered histone acetylation.

Various enteroinvasive and pathogenic bacteria induce a complex profile of proinflammatory genes in intestinal epithe-lial cells (IEC) 1 by activating multiple signaling cascades, including the transcription factor NF-B (1,2). More recently, we have shown that the non-pathogenic commensal enteric bacterial species Bacteroides vulgatus activates ICAM-1 gene expression through a mechanism involving both NF-B nuclear translocation and RelA phosphorylation in primary and IEC lines (3). These data suggest that non-pathogenic commensal as well as pathogenic bacteria have the potential to initiate innate host responses in IEC. The key role for commensal luminal bacteria in the pathogenesis of inflammatory bowel disease (IBD) is supported by the compelling evidence that intestinal inflammation is dramatically reduced or absent in germ-free rodent models susceptible for the development of chronic experimental colitis (4 -7). In addition, reconstitution studies of gnotobiotic HLA-B27 transgenic rats (8,9) and carrageenan-induced colitis in guinea pigs (10) suggest that selected bacterial species such as B. vulgatus are particularly important to the induction of colitis in these models. Of note, these bacteria induced no colitis in wild type rats, documenting their non-pathogenic nature (8,9). We hypothesize that commensal non-pathogenic bacteria-induced NF-B transcriptional activity and gene expression is regulated by a complex network of anti-inflammatory molecules and regulatory immune cells that prevents chronic intestinal inflammation in a normal host. Understanding the molecular mechanisms of IEC hyporesponsiveness to constant bacterial exposure would reveal important regulatory elements involved in the maintenance of intestinal homeostasis.
IEC must adapt to a constant and constitutively changing luminal environment by processing different biological information through multiple signaling cascades that target a defined set of genes, which then provide an adequate effector response (2). Importantly, these extracellular cytoplasmic signals must be integrated with the action of various nuclear signal processes controlling access of transcription factors to the regulatory region of gene promoters. In the case of NF-B, this implies induction of a signaling cascade that leads to the release of the transcription factor from its cytoplasmic inhibitor IB, shuttling of the transcription factor to the nucleus where it accumulates. Simultaneously, activation of nuclear signaling proteins permits access to gene promoter regions (11). In the resting state, transcriptionally inactive chromatin is tightly wrapped around histone proteins that hinder DNA binding by transcriptional factors, providing an additional complex layer of gene regulation. Histone wrapping to DNA and consequently accessibility of transcription factors to gene promoter elements is regulated by post-translational modifying mechanisms, such as histone acetylation, phosphorylation, and methylation. Signaling cascades responsible for the activation of transcription factors must act synchronously with the different pathways leading to chromatin unfolding in order to achieve coordinate gene expression. For example, signaling cascades that impact the levels of histone acetylation strongly influence chromatin remodeling and gene expression (11)(12)(13). The status of histone acetylation is dictated by the opposite action of the two enzymes histone acetyltransferase (HAT), which catalyzes acetylation of lysine residues present at the N-terminal part of the histone tail, and the histone deacetyltransferase (HDAC), which removes acetyl groups. Indeed, NF-B binding to transcriptional active chromatin is influenced by the status of histone acetylation/phosphorylation (14 -16). Whether the level of histone acetylation impacts bacteria-induced NF-B-dependent gene expression in IEC is currently unknown.
In vivo, the response of IEC to commensal bacteria is regulated by adjacent lamina propria mononuclear cells, which mediate immunologic tolerance and maintain mucosal homeostasis (17). Among the anti-inflammatory molecules implicated in the regulation of intestinal homeostasis, transforming growth factor (TGF)-␤1 has been shown to play a key role in controlling proliferation, differentiation, and function of numerous immune and non-immune cells (18 -20). The biological effect of TGF-␤1 is mediated through activation of various signaling cascades such as Smad and the mitogen-activated protein kinases, the latter including extracellular signal-regulated kinases, stress-activated protein kinases/c-Jun N-terminal kinases, and the p38 pathway. Upon TGF-␤1 binding to the receptor complex, the Smad signaling pathway is activated through phosphorylation of the receptor activated Smad1-3 (R-Smads), which dissociate from the receptor complex and associate with the common partner Smad4 (co-Smad4). Resulting heterodimers then translocate to the nucleus and directly bind to Smad-binding elements on the DNA to induce target gene transcription (21)(22)(23)(24). Another important aspect of TGF-␤ signal transduction is the induction/activation of inhibitory molecules such as Smad6 and Smad7, which act as negative regulators of TGF-␤1/Smad signaling by interfering with either the activation of R-Smads or the association of R-Smads with Smad4 (25).
Although TGF-␤1 regulates the host immune response to various micro-organisms and inflammatory mediators (26,27), the molecular mechanisms remain to be elucidated. In this study, we characterized TGF-␤1 signal transduction in IEC and investigated its effect on bacteria-induced NF-B signaling. We report that TGF-␤1-activated Smad signaling mediates inhibition of B. vulgatus and LPS-induced NF-B recruitment to the IL-6 gene promoter through modulation of histone acetylation. Activation of this inhibitory pathway may be important in down-regulating IEC responses to commensal bacteria.
Immunohistochemistry and Isolation of Primary Intestinal Epithelial Cells-Germ-free (sterile) Fisher F344 rats were monoassociated at 10 -12 weeks of age with B. vulgatus and maintained in the Gnotobiotic Animal Core at the College of Veterinary Medicine, North Carolina State University, Raleigh. Bacterial monoassociation was confirmed by anaerobic culture of stool samples. Rats were killed at day 3 and 7 after initial bacterial colonization. Sections of the ileum, cecum, and proximal and distal colon were fixed in 10% neutral buffered formalin. The fixed tissue was embedded in paraffin. Immunohistochemistry was performed using anti-phospho-NF-B p65 (RelA), anti-phospho-Smad2, anti-phospho-c-Jun, and anti-phospho-p38 Abs (Cell Signaling, Beverly, MA) according to the protocol of the manufacturer, and sections were counterstained with hematoxylin. Histology scoring was analyzed by scoring the degree of lamina propria mononuclear cell infiltration, crypt hyperplasia, goblet cell depletion, and architectural distortion as described previously (8).
Adenoviral Infection-CMT93 cells were infected overnight with Ad5B-LUC as described previously (3). Where indicated CMT93 were co-infected for an additional 12 h with Ad5IB␣AA or Ad5Smad7 (a generous gift from Dr. David A. Brenner, University of North Carolina, Chapel Hill, NC) or Ad5Smad3⌬C or Ad5Smad3/Ad5Smad4 (a generous gift of Dr. X. F. Wang, Duke University, Durham, NC) at a multiplicity of infection (m.o.i.) of 50. The Smad3⌬C, Smad3, and Smad4 constructs were described previously by Wu et al. (28) and Zhang et al. (29). The Ad5GFP containing green fluorescent protein was used as a viral negative control. The adenoviruses were washed off, and fresh medium containing serum was added. Cells were stimulated at various time points with B. vulgatus (5 ϫ 10 7 cfu/ml), B. vulgatus lysate (100 g/ml), or LPS (10 g/ml).
Transfection and Reporter Gene Assay-CMT93 cells were transfected with plasmid expressing 3TP-LUX (1 g, a generous gift of Dr. Richard A. Rippe, University of North Carolina, Chapel Hill) using LipofectAMINE Reagent (Invitrogen) as described previously (3). Cells were then stimulated with TGF-␤1 (20 ng/ml) for 12 h. Cell extracts were prepared using enhanced luciferase assay reagents (Analytical Luminescence, San Diego). Luciferase assay were performed on a Monolight 2010 luminometer for 20 s (Analytical Luminescence, San Diego), and results were normalized for extract protein concentrations measured with the Bio-Rad protein assay kit (Bio-Rad).
Chromatin Immunoprecipitation (ChIP) Analysis-CMT93 cells were stimulated with LPS for 0 -2 h. After stimulation, IEC were washed in cold phosphate-buffered saline (1 time) and fixed by adding formaldehyde to a final concentration of 1%. Nuclear extraction and chromatin immunoprecipitation were performed as described previously (16). Cells were lysed after formaldehyde fixation in L1 lysis buffer (50 mM Tris, pH 8.0, 2 mM EDTA, 0.1% Nonidet P-40, and 10% glycerol) supplemented with protease inhibitors. Nuclei were pelleted and resuspended in 300 l of L2 lysis buffer (50 mM Tris, pH 8.0, 0.1% SDS, and 5 mM EDTA). Chromatin was sheared by sonication (4 times for 15 s at about one-fifth of maximum power), centrifuged, and diluted in dilution buffer (50 mM Tris, pH 8.0, 0.5% Nonidet P-40, 0.2 M NaCl, and 0.5 mM EDTA). Extracts were normalized according to their DNA concentration and were precleared for 3 h with salmon sperm-saturated protein A/G-agarose. Immunoprecipitation were carried out overnight at 4°C using 10 l of anti-NF-B p65 Ab (Rockland, Gilbertsville, PA). Immune complexes were collected with salmon sperm-saturated protein A/G-agarose for 30 min and washed three times in high salt buffer (20 mM Tris, pH 8.0, 0.1% SDS, 0.5 M NaCl, 1% Nonidet P-40, 2 mM EDTA) followed by three washes with no salt buffer (TE 1ϫ). Immune complexes were extracted 3 times with 150 l of extraction buffer (1% SDS, 0.1 M NaHCO 2 ). DNA cross-links were reverted by heating for 8 h at 65°C. After proteinase K (100 g for 1 h) digestion, DNA was extracted with phenol/chloroform and precipitated in ethanol. DNA isolated from an aliquot of the total nuclear extract was used as a loading control for the PCR (input control). PCR was performed with total DNA (1 l, input control) and immunoprecipitated DNA (1 l) using the following IL-6 promoter-specific primers: IL-6A (5Ј) 5-GA-CATGCTCAAGTGCTGAGTCAC-3 (position 987-1009) and IL-6B (3Ј) 5-AGATTGCACAATGTGACGTCG-3 (position 1112-1132). The length of the amplified product was 125 bp. The PCR products (10 l) were subjected to electrophoresis on 2% agarose gels containing gel Star fluorescent dye (FMC, Philadelphia, PA). Fluorescent staining was captured using an AlphaImager 2000 (AlphaInnotech, San Leandro, CA).
In Vitro Kinase Assays-IKK activity on B. vulgatus-induced serine RelA phosphorylation was determined by immunocomplex kinase assay as described previously (3). Briefly, CMT93 cells were lysed in Triton lysis buffer containing protease and phosphatase inhibitors after stimulation with B. vulgatus/LPS at various times in the presence or absence of TGF-␤1. 300 g was immunoprecipitated with 2 l of anti-IKK␥ (Cell Signaling, Beverly, MA), and the kinase reaction was performed by glutathione S-transferase (GST)-p65 (1-305 and 354 -551) (a generous gift of Dr. Hiroaki Sakurai, Tanabe Seiyaku, Osaka, Japan) as substrate for 30 min at 30°C. Substrate protein was resolved by gel electrophoresis, and phosphate incorporation was assessed by audioradiography and PhosphorImager analysis (Amersham Biosciences).
Co-immunoprecipitation-CMT93 cells were stimulated with B. vulgatus/LPS for 30 min in the presence or absence of TGF-␤1. Where indicated CMT-93 cells were co-infected with Ad5Smad3 and Ad5Smad4. Nuclear extract were prepared as described above, and protein concentration was normalized. Immunoprecipitation were carried out overnight at 4°C using 10 l of anti-NF-B p65 Ab (Rockland, Gilbertsville, PA). Immune complexes were collected with protein A/Gagarose for 30 min and washed 3 times in high salt buffer (20 mM Tris, pH 8.0, 0.1% SDS, 0.5 M NaCl, 1% Nonidet P-40, 2 mM EDTA) followed by three washes with no salt buffer (1 TE). The immune complexes were cleaved from the beads by adding 50 l of 1ϫ Laemmli buffer. Western blot analysis was carried out using anti-CBP/p300 Ab (Upstate, Lake Placid, NY) as described above.
Nuclear Extracts and Electrophoretic Mobility Shift Assay-CMT93 cells were stimulated for various times (0 -4 h) with B. vulgatus lysate (100 g/ml), and nuclear extracts were prepared as described previously (3). Extracts (5 g) were incubated with radiolabeled double-stranded class I major histocompatibility complex B sites (GGCTGGGGATTC-CCCATCT), separated by nondenaturating electrophoresis, and analyzed by autoradiography as described previously (30).
ELISA-CMT93 cells were stimulated with B. vulgatus and LPS for 16 -24 h. Murine IL-6 was determined in the cell culture supernatant using ELISA (R & D Systems).
Statistical Analysis-Data are expressed as means Ϯ S.D. of triplicate experiments. Statistical significance was performed by the twotailed Student's t test for paired data and considered significant if p values were Ͻ0.05 and Ͻ0.01.

B. vulgatus Monoassociated Rats Show Nuclear
Localization of Phospho-RelA and Phospho-Smad2 in IEC-We first sought to demonstrate activation of NF-B and TGF-␤ signaling in vivo following B. vulgatus monoassociation of germ-free Fisher F344 rats. Monoassociated rats were killed 3 and 7 days after initial bacterial colonization, and sections of the ileum, cecum, proximal and distal colon were analyzed by using immunohistochemical analysis and histological scoring. Fig. 1A shows nuclear localization of phospho-RelA (Ser-536) in cecal epithelial cells 3 days after B. vulgatus monoassociation but not at day 7 or in germ-free controls, suggesting transient NF-B activation in intestinal epithelial cells after initial bacterial colonization. Nuclear phospho-RelA staining was stronger in the surface region of the cecal epithelium and almost absent in the ileum, the proximal and distal colon, or in lamina propria cells (Fig. 1B). None of the histologically examined sections Germfree Fisher F344 rats were monoassociated with B. vulgatus. Rats were killed at days 3 and 7 (3d and 7d) after initial bacterial colonization. Sections of the ileum, cecum, and proximal and distal cecum were fixed in 10% neutral buffered formalin. Immunohistochemistry was performed on paraffin-embedded tissue sections using anti-RelA antibody, and sections were counterstained with hematoxylin. A, nuclear phospho-RelA staining of cecal sections from germ-free and bacterial monoassociated rats is shown at a magnification of ϫ20. Isotype antibodies are used as a control. B, nuclear phospho-RelA staining of sections from ileum, cecum, and proximal and distal colon of B. vulgatus-monoassociated rats (3 days) is shown at a magnification of ϫ40. Representative sections of two different experiments are shown. revealed signs of pathologic inflammation.
Interestingly and in contrast to RelA, nuclear phospho-Smad2 is sustained throughout days 3 and 7 after bacterial monoassociation, with no signal detected in germ-free controls ( Fig. 2A). Nuclear phospho-Smad2 staining was first detected in the lower crypt region of the cecal epithelium at day 3 but was strongest in the surface region at day 7 after initial bacterial colonization of germ-free rats. Similar to the localization of phospho-RelA, phospho-Smad2 was almost absent in the ileum, the proximal and distal colon, or in lamina propria cells (Fig. 2B). In addition, bacteria-induced RelA and Smad2 phosphorylation in primary large intestinal epithelial cells isolated from B. vulgatus-monoassociated germ-free rats was confirmed by Western blot analysis (Fig. 3A). We next sought to determine whether B. vulgatus induces RelA and Smad2 phosphorylation ex vivo in primary IEC isolated from germ-free rats. Fig. 3B shows that B. vulgatus stimulation induces RelA but not Smad2 phosphorylation, suggesting that bacteria-induced Smad signaling is mediated by activation of TGF-␤ production by adjacent intestinal lamina propria mononuclear cells.
These data suggest different temporal and spatial phospho-RelA and phospho-Smad2 distribution in B. vulgatus-monoassociated rats and demonstrate that TGF-␤1 and NF-B signaling pathways are activated in the intestine in vivo after nonpathogenic bacterial colonization.
TGF-␤1 Activates the Smad Signal Transduction Pathway in IEC Lines-Since B. vulgatus-monoassociated rats showed nuclear localization of phospho-Smad2 in IEC (Figs. 1 and 2), we   FIG. 2. A and B, nuclear localization of phospho-Smad2 in the intestinal epithelium of B. vulgatus-monoassociated rats. Germfree Fisher F344 rats were monoassociated with B. vulgatus. Rats were killed at day 3 and 7 (3d and 7d) after initial bacterial colonization. Sections of the ileum, cecum, proximal and distal cecum were fixed in 10% neutral buffered formalin. Immunohistochemistry was performed on paraffin-embedded tissue sections using anti-Smad2 antibody, and sections were counterstained with hematoxylin. A, nuclear phospho-Smad2 staining of cecal sections from germ-free and B. vulgatus-monoassociated rats (days 3 and 7) is shown at a magnification of ϫ20. Isotype antibodies are used as a control. B, nuclear phospho-Smad2 staining of sections from ileum, cecum, and proximal and distal colon of B. vulgatus-monoassociated rats (7 days) is shown at a magnification of ϫ40. Representative sections of two different experiments are shown. FIG. 3. A and B, B. vulgatus induces RelA but not Smad2 phosphorylation in isolated primary IEC. A, germ-free Fisher F344 rats were monoassociated with B. vulgatus. Rats were killed at day 3 and 7 after initial bacterial colonization and primary IEC from the cecum and colon were isolated. Total protein was extracted, and 20 g of protein was subjected to SDS-PAGE followed by phospho-RelA, phospho-Smad2, and ␤-actin immunoblotting using the ECL technique. B, primary IEC from cecum and colon were isolated from germ-free Fisher F344 rats. IEC were seeded into 6-well plates at a density of 2 ϫ 10 6 cells per ml and stimulated with B. vulgatus (7 ϫ 10 7 cfu/ml) for 0 -4 h. Total protein was extracted, and 20 g of protein was subjected to SDS-PAGE followed by phospho-RelA, phospho-Smad2, and ␤-actin immunoblotting using the ECL technique.
next sought to investigate TGF-␤1 signal transduction in IEC. It is well established that TGF-␤1 activates the Smad signaling pathway in numerous cell types (31); however, TGF-␤1 signaling has not yet been investigated in IEC. TGF-␤1 induced rapid (15 min) Smad2 phosphorylation in the colonic mouse epithelial cell line CMT93, whereas the level of total unphosphorylated protein remained unaffected (Fig. 4A). We next investigated whether phosphorylated Smad2/3 associates with co-Smad4 in TGF-␤1-stimulated CMT93 cells. As shown by immunoprecipitation analysis, phosphorylated Smad2 and Smad3 (Fig. 4B, upper panel) rapidly associated with Smad4 in TGF-␤1-stimulated CMT93 cells, but this association was transient with little evidence of complexing at 60 min. TGF-␤1activated Smad2 phosphorylation was confirmed in mouse pri-mary IEC isolated from 129SvEv wild type mice (Fig. 4C). These results suggest that TGF-␤1 induces the Smad signaling cascade in primary and IEC lines. Finally, we investigated the functional consequences of TGF-␤1-induced Smad activation in CMT93 cells by using LipofectAMINE-based transfection of the TGF-␤-responsive reporter gene 3TP-LUX. Fig. 4D shows an almost 4-fold increase in luciferase activity in TGF-␤1-stimulated CMT93 cells.
TGF . Similar results were obtained in the rat small intestinal epithelial cell line IEC-6 (data not shown). These results demonstrate that TGF-␤1 significantly, but incompletely, inhibits Gram-negative bacteria-induced NF-B transcriptional activity in IEC.
Next, we evaluated whether TGF-␤1 inhibits the expression of endogenous B-dependent genes. We stimulated CMT93 cells with B. vulgatus or LPS in the presence or absence of TGF-␤1 and measured IL-6 mRNA expression as well as protein secretion after 6 and 24 h, respectively. B. vulgatus and LPS-induced IL-6 mRNA expression (Fig. 5B) and protein secretion (Fig. 5C) is significantly inhibited by TGF-␤1 treatment.
We have previously shown that B. vulgatus/LPS-induced NF-B transcriptional activity requires IB␣ and RelA phosphorylation as well as activation of phosphatidylinositol 3-kinase/Akt pathway (3). Based on this finding, we first investigated the effect of TGF-␤1 on B. vulgatus/LPS-induced NF-B signal transduction by measuring endogenous IB␣, RelA, and Akt phosphorylation as well as NF-B DNA-binding in CMT93 cells. Interestingly, TGF-␤1 did not alter B. vulgatus/LPS-induced IB␣, RelA, and Akt phosphorylation (Fig. 6A) nor NF-B DNA binding (Fig. 6C). Moreover, B. vulgatus/LPSinduced RelA nuclear translocation as determined by immunofluorescence was not inhibited by TGF-␤1 treatment (data not shown).
B. vulgatus induces RelA phosphorylation in IEC at both serine residues 536 and 529 (3). Since our antibody selectively recognizes phosphorylated serine 536, we next used an in vitro kinase assay to ascertain the effect of TGF-␤1 on bacteriainduced RelA phosphorylation. We immunoprecipitated IKK␥ and performed a kinase assay using a GST fusion protein containing the p65 transactivating domain (p65TAD; amino acids 354 -551) as described previously (3). Fig. 6B demonstrates that GST-p65TAD (354 -551) is phosphorylated in LPS stimulated CMT93 cells with a kinetic pattern similar to endogenous RelA Ser-536 phosphorylation (Fig. 6A). TGF-␤1 treatment did not affect LPS stimulated GST-p65TAD (354 -551) phosphorylation, suggesting that the inhibitory effect of this cytokine operates independently of signal-induced proximal NF-B signaling.

FIG. 4. A-D, TGF-␤1 activates the Smad signal transduction pathway in primary colonic and IEC lines.
A, CMT-93 cells were incubated with TGF-␤1 (20 ng/ml) for 15 min to 1 h. Total protein was extracted, and 20 g of protein was subjected to SDS-PAGE followed by phospho-Smad2 and Smad2 immunoblotting using the ECL technique. B, CMT93 cells were stimulated with TGF-␤1 (20 ng/ml) for 30 min and 1 h. 300 g of protein were immunoprecipitated using anti-Smad4 Ab. Western blot analysis was performed by subjecting 20 g of immunoprecipitated protein to SDS-PAGE followed by phospho-Smad2/3 and Smad4 immunoblotting using the ECL technique. C, primary IEC were isolated from 129SvEv wild type mice as described under "Materials and Methods." Primary cells were stimulated for 30 min with TGF-␤1 (20 ng/ml) or medium alone. Total protein was extracted, and 20 g was subjected to SDS-PAGE followed by phospho-Smad2 and ␤-actin immunoblotting using the ECL technique. D, CMT93 cells were transfected with 3TP-LUX reporter gene (1 g) and stimulated with TGF-␤1 or medium alone for 12 h. Cell extracts and the luciferase assay were performed as described under "Materials and Methods," and results were normalized for extract protein concentrations. These results are representative of two independent experiments. Ad5Smad7 and AdSmad⌬3 Reverse the Inhibitory Effect of TGF-␤1 on Bacteria-induced NF-B Transcriptional Activity-Proximal TGF-␤1 signal transduction has been shown to be down-regulated by the endogenous inhibitor Smad7 (31). To study the functional role of TGF-␤1/Smad in blocking B. vulgatus-induced NF-B transcriptional activity, we used an adenoviral vector to deliver biologically active Smad7 (Ad5S-mad7). Initially, CMT93 cells were infected with Ad5B-LUC in combination with Ad5Smad7 and then stimulated with B. vulgatus or LPS. Interestingly, TGF-␤1-mediated inhibition of B. vulgatus- (Fig. 7A) and LPS-induced (Fig. 7B) NF-B transcriptional activity was completely reversed in Ad5Smad7-but not Ad5GFP-infected cells. Infection of CMT93 cells with Ad5Smad7 was confirmed by Western blot analysis using anti-Smad7 and anti-HA Abs (Fig. 7A).
To document further the involvement of Smad signaling in TGF␤-induced inhibitory effect on NF-B transcriptional activity, we used a dominant negative Smad3 (Ad5Smad⌬3). Similarly to Smad7, TGF-␤1-induced suppression of B. vulgatusand LPS-induced NF-B transcriptional activity is inhibited in Ad5Smad⌬3-infected CMT-93 cells (Fig. 7, C and D). Infection of CMT93 cells with Ad5Smad⌬3 was confirmed by Western blot analysis using anti-HA Abs (Fig. 7C).

TGF-␤1 Inhibits Histone 3 Acetylation/Phosphorylation and Bacteria-induced Recruitment of RelA to the IL-6 Promoter-
Previous reports (14 -16) have demonstrated that changes in histone acetylation/phosphorylation strongly influence recruitment of NF-B to selected target genes. Because TGF-␤1-mediated inhibition of IL-6 gene expression occurs independently of the classical NF-B signal transduction, we next investigated the effect of this immunosuppressive cytokine on histone acetylation levels. We first used the specific histone deacetylase inhibitor TSA to determine whether the inhibitory effect of TGF-␤1 is dependent of the status of histone acetylation. Interestingly, B. vulgatus-and LPS-induced histone 3 acetylation/phosphorylation (Lys-9/Ser-10) is inhibited in TGF-␤1treated cells (Fig. 8A, top and middle panels) but reversed in the presence of TSA (Fig. 8A, lower panels). Moreover, TGF-␤1-mediated inhibition of LPS-induced IL-6 secretion was reversed in TSA-treated cells (Fig. 8B). This suggests that TGF-␤1-mediated inhibition of IL-6 gene expression involved changes in histone acetylation.
To address specifically the effect of TGF-␤1-mediated repression of B. vulgatus/LPS-induced IL-6 gene expression, we performed ChIP using a p65 antibody. CMT93 cells were stimulated with LPS in the presence or absence of TGF-␤1, and p65 recruitment to the IL-6 promoter region was determined by PCR. Fig. 8C shows that LPS strongly induced p65 recruitment to the IL-6 promoter at 2 h (Fig. 8C, upper panel), which was blocked by TGF-␤1 treatment (Fig. 8C, upper panel). Similar levels of IL-6 were amplified from the total pool of genomic DNA (input control), documenting that equal amounts were used for the ChIP analysis (Fig. 8C, lower panel). To address the role of Smad signaling on TGF-␤1-mediated inhibition of IL-6 gene expression, we next used Ad5Smad7 or Ad5GFPinfected cells for ChIP analysis. Ad5Smad7 but not Ad5GFP reversed TGF-␤1-mediated inhibition of NF-B recruitment to the IL-6 gene promoter (Fig. 8D). This suggests that TGF-␤1induced Smad signaling is necessary for inhibition of B. vulgatus/LPS-induced RelA recruitment to the IL-6 promoter in IEC.  5. A-C, TGF-␤1-mediated inhibition of B. vulgatus-induced NF-B transcriptional activity and IL-6 gene expression in IEC. A, CMT93 cells were infected with Ad5B-LUC (5 m.o.i.) for 12 h. Cells were washed and co-infected for additional 12 h with Ad5IB␣AA. Finally, IEC were stimulated with B. vulgatus (5 ϫ 10 7 cfu/ml), B. vulgatus lysate (100 g total protein/ml), LPS (10 g/ml), or medium alone for additional 12 h in the presence or absence of TGF-␤1 (20 ng/ml). Cell extracts and the luciferase assay were performed as described under "Materials and Methods," and results were normalized for extract protein concentrations. B, CMT93 cells were stimulated with B. vulgatus lysate and LPS for 6 h in the presence or absence of TGF-␤1 (20 ng/ml). Total RNA was extracted, reverse-transcribed, and amplified using specific IL-6 and GAPDH primers. PCR products were run on a 2% agarose gel and stained with gel star. These results are representative of two independent experiments. C, CMT93 cells were stimulated with B. vulgatus lysate and LPS, and IL-6 protein secretion was measured in the cell culture supernatants after 24 h of stimulation by p300 has intrinsic HAT activity and associates with NF-B to induce transcription (32,33), we next investigated the effect of TGF-␤1 or Ad5Smad3 and Ad5Smad4 on p65 and CBP/p300 association. Cells were treated with TGF-␤1 or Ad5Smad3/ Ad5Smad4 and then stimulated with LPS. Interestingly, coimmunoprecipitation experiment demonstrates that TGF-␤1 treatment reversed LPS-induced nuclear CBP/p300 association with p65 (Fig. 9, compare lanes 2 with 4). Moreover, the presence of Ad5Smad3/Ad5Smad4 also reversed LPS-induced CBP/ p300 association with p65 (Fig. 9, compare lanes 2 and 5), suggesting that TGF-␤1-induced Smad signaling may prevent bacteria-induced p65-CBP/p300 interaction in IEC.

TGF-␤1 Inhibits B. vulgatus-induced CBP/p300 and RelA
In conclusion, these results suggest that B. vulgatus/LPSinduced IL-6 gene expression in IEC is partially regulated by TGF-␤1 through decreased histone acetylation/phosphorylation and loading of NF-B to the gene promoter.

DISCUSSION
Numerous reports point to a critical immunoregulatory function of TGF-␤1 in the maintenance of intestinal homeostasis. First, Yang et al. (34) showed that targeted disruption of the Smad3 gene caused massive T cell infiltration and pyogenic abscess formation in the stomach and intestine. Subsequently, Monteleone et al. (35) showed that Smad7, the endogenous inhibitor of the Smad signal transduction pathway, is overexpressed in mucosal T cells from patients with IBD. In vitro blocking of Smad7 restored TGF-␤1-mediated Smad phosphorylation and enhanced TGF-␤1-mediated inhibition of inflammatory cytokine production in lamina propria mononuclear cells isolated from IBD patients. Moreover, TGF-␤1-deficient mice spontaneously develop colitis (36), and overexpression of TGF-␤1 in lamina propria immune cells inhibited Th1-mediated experimental 2,4,6-trinitrobenzene sulfonic acid-induced colitis (37,38). The absence of colitis and pathogenic immune responses to commensal enteric bacteria suggests that normal hosts have well developed immunosuppressive mechanisms of their mucosal immune systems.
Although the immunosuppressive effect of TGF-␤1 is well established, the molecular mechanisms responsible for the immunomodulation remain to be elucidated. We now demonstrate that TGF-␤1/Smad signaling inhibits B. vulgatus and LPSinduced IL-6 gene expression through reduction of histone acetylation/phosphorylation and decreased recruitment of NF-B to the IL-6 promoter. First, we show for the first time a sequential increase of nuclear RelA (day 3) and Smad2 (days 3 and 7) phosphorylation in cecal epithelial cells of B. vulgatusmonoassociated germ-free Fisher F344 rats, suggesting that this commensal bacterial strain induced both pro-inflammatory and immunosuppressive signaling pathways in the host in vivo. Second, by using primary and colonic cell lines, we provide evidence that TGF-␤1 induced Smad2 phosphorylation, association of phospho-Smad2/Smad3 with Smad4, and enhanced TGF-␤1-dependent transcriptional activity, suggesting a functional TGF-␤1-mediated Smad signal transduction in IEC. Third, we show that TGF-␤1 significantly inhibits B. vulgatus and LPS-induced NF-B transcriptional activity as well as IL-6 gene expression in IEC, demonstrating the biological significance of TGF-␤1 signal transduction in regulating IEC responsiveness to luminal Gram-negative bacteria and bacterial products.
We have shown previously that B. vulgatus-induced NF-B activation and pro-inflammatory gene expression require en-LPS (10 g/ml) for 2 h in the presence or absence of TGF-␤1 (20 ng/ml). NF-B DNA-binding activity was determined using electrophoretic mobility shift assay. These results are representative of two independent experiments.
FIG . 6. A-C, role of TGF-␤1 on B. vulgatus-induced NF-B signaling cascade. A, CMT93 cells were stimulated with B. vulgatus lysate (100 g of protein/ml) for 0 -2 h in the presence or absence of TGF-␤1 (20 ng/ml). Where indicated CMT93 cells were pretreated with 20 M MG132. Total protein was extracted, and 20 g of protein was subjected to SDS-PAGE followed by phospho-IB␣, RelA, and Akt immunoblotting using the ECL technique. B, LPS-induced serine RelA phosphorylation in the presence or absence of TGF-␤1 (20 ng/ml) was determined by an immunocomplex kinase assay as described under "Materials and Methods." The kinase reaction was performed by incubating 25 l of kinase buffer with either glutathione S-transferase GST-p65 (residues 354 -551) as substrate for 30 min at 30°C. Substrate protein was resolved by gel electrophoresis, and phosphate incorporation was assessed by audioradiography and PhosphorImager analysis. Coomassie Blue staining shows equal loading (lower blot). C, CMT93 cells were stimulated with B. vulgatus lysate (100 g of protein/ml) or hanced phosphatidylinositol 3-kinase/Akt and RelA phosphorylation (3). However, TGF-␤1 did not affect B. vulgatus-induced IB␣, RelA, and Akt phosphorylation nor alter NF-B DNA-binding, suggesting an alternative mechanism for the TGF-␤1-mediated inhibition of B. vulgatus-induced NF-B transcriptional activity in IEC. Of interest, levels of histone acetylation/phosphorylation directly impact expression of some NF-B-dependent genes by controlling access of the transcription to the appropriate gene promoter (39). In addition, recent studies (40,41) demonstrate that TGF-␤/Smad-mediated transcriptional repression involves recruitment of HDAC. Interestingly, we demonstrate that the HDAC inhibitor TSA reversed the TGF-␤1-mediated inhibitory effect on B. vulgatus/LPS-induced IL-6 gene expression, suggesting that TGF-␤1 modulates the status of histone acetylation. More directly, we show that B. vulgatus/LPS-induced histone acetylation/phosphorylation is inhibited by TGF-␤1 treatment. Similarly, we show that loading of NF-B to the IL-6 promoter is inhibited by TGF-␤1 treatment in IEC. Together, these data point to a new regulatory mechanism of TGF-␤1-mediated immunosuppression which involves decreased histone acetylation and inhibition of NF-B accession to the gene promoter.
The mechanisms controlling the coordinated nuclear migration of transcription factors and access to various gene promoters are not elucidated. Recently, activation of the mitogenactivated protein kinase p38 was shown to be critical in LPSmediated histone phosphorylation and NF-B access to select gene promoters in dendritic cells (16). Interestingly, we failed to observe enhanced p38 phosphorylation in B. vulgatus-or LPS-stimulated CMT93 cells. 2 Although a role for p38 in bacteria and LPS-induced histone acetylation/phosphorylation could not be totally ruled out, our data suggest that other pathways may participate in histone modification, at least in IEC.
The precise mechanism of enhanced NF-B transcriptional activity by signal-induced RelA phosphorylation is not clearly established. Zhong et al. (32,33) have shown that LPS-induced RelA serine 276 phosphorylation loosens its interaction with HDAC and promotes its binding to the transcriptional co-activator CBP/p300 containing HAT activity. Interestingly, association of HDAC1 and HDAC2 with unphosphorylated nuclear RelA is associated with transcriptionally inactive NF-B (42). However, we found no correlation between TGF␤1-mediated inhibition of RelA recruitment to the IL-6 gene promoter and the level of RelA phosphorylation. This was confirmed by investigating both endogenous serine 536 phosphorylation levels by Western blot and by an in vitro kinase assay using a GST-RelA substrate (serine 526 and 529). Since LPS-induced serine 276 phosphorylation strongly modulates NF-B transcriptional activity (32,33), the possibility remains that TGF-␤1 interferes with this phosphorylation site and prevents bacteria-induced NF-B transcriptional activity. However, RelA serine 276 phos-  7. A-D, TGF-␤1-mediated inhibition of B. vulgatus/LPSinduced NF-B transcriptional activity was reversed by Ad5Smad7 and Ad5Smad⌬3. CMT93 cells were infected with Ad5B-phorylation is not induced in B. vulgatus-and LPS-stimulated IEC (3), suggesting that the TGF-␤1-mediated inhibition operates independently of the RelA phosphorylation status. Therefore, although we observed a decrease in RelA phosphorylation (day 7) but enhanced Smad phosphorylation (days 3-7) in B. vulgatus monoassociated rats, these two events are likely independent of each other. Decreased RelA phosphorylation may represent an intrinsic down-regulatory mechanism present in IEC.
B. vulgatus signaling to RelA phosphorylation and NF-B DNA binding activity is intact in TGF␤1-treated cells, but RelA recruitment to the IL-6 promoter is blocked. This suggests that the immunoregulatory cytokine interferes with the activity of a co-repressor or co-activator of NF-B. Interestingly, TGF-␤1mediated Smad signal transduction is necessary for inhibition of B. vulgatus/LPS-induced NF-B transcriptional activity and IL-6 gene expression, as demonstrated by reversed suppression in Ad5Smad7 as well as in Ad5Smad⌬3-infected cells. Of note, TGF-induced nuclear Smad complex associates with the CBP/ p300 transcriptional co-activator (43)(44)(45). Interestingly, we show that TGF-1 and the presence of Ad5Smad3/Ad5Smad4 reversed LPS-induced CBP/p300 p65 nuclear interaction. This suggests that the mechanism of TGF-␤-induced inhibitory effect involved decreased p65-CBP/p300 association, possibly through negative titration by Smad proteins. Therefore, diminished CBP/p300 association with p65 in conjunction with decreased HAT activity could lead to impaired histone acetylation/phosphorylation and/or lower p65 recruitment to the IL-6 gene promoter. The formal interaction of Smad protein with CBP/p300 and the precise role of this co-activator in bacteria signaling in IEC will necessitate further investigation.
In conclusion, we demonstrate that TGF-␤1 inhibits non-  8. A-D, TGF-␤1 inhibits B. vulgatus/LPS-induced recruitment of NF-B to the IL-6 promoter through blockade of histone acetylation. A, CMT93 cells were stimulated with B. vulgatus lysate (100 g of protein/ml) and LPS (10 g/ml) for 0 -2 h in the presence or absence of TGF-␤1 (20 ng/ml). Where indicated cells were pretreated with trichostatin A (2 h pretreatment with 1 g/ml). Total protein was extracted, and 20 g of protein was subjected to SDS-PAGE followed by acetyl-phospho-histone 3 (Lys-9/Ser-10) and ␤-actin immunoblotting using the ECL technique. B, CMT93 were stimulated with LPS for 16 h in the presence or absence of TGF-␤1. Where indicated cells were pretreated with trichostatin A (2 h pretreatment with 1 g/ml). IL-6 FIG. 9. TGF-␤1 and Ad5Smad3/Ad5Smad4 inhibits nuclear association of CBP/p300 with p65. Where indicated cells were coinfected with Ad5Smad3/Ad5Smad4 overnight. CMT-93 cells were stimulated with LPS (10 g/ml) for 30 min in the presence or absence of TGF-␤1 (20 ng/ml). Cells were lysed, and the nuclei were pelleted and resuspended in lysis buffer. Extracts were normalized according to their protein concentration. Immunoprecipitations (IP) were carried out overnight using 10 l of anti-NF-B p65 Ab. Immune complexes were collected with protein A/G-agarose and washed three times in high salt followed by three washes with no salt buffer. Western blot (WB) analysis was carried out using anti-CBP/p300 Ab as described under "Materials and Methods." pathogenic commensal Gram-negative bacteria-induced histone acetylation/phosphorylation and RelA loading to the IL-6 gene promoter in IEC. We suggest that the immunoregulatory cytokine TGF-␤1 participates in the maintenance of intestinal homeostasis by decreasing the ability of commensal luminal enteric bacteria to activate the NF-B signaling pathway at the level of gene promoter accessibility.