Enterococcus faecalis from newborn babies regulate endogenous PPARγ activity and IL-10 levels in colonic epithelial cells

Synthetic Ligands Induce Phosphorylation of Endogenous PPARγ1 in HT-29.

To assess the impact of ligands on the induction of PPARγ phosphorylation, we monitored the protein phosphorylation status in the colonic adenocarcinoma cell line HT-29. Cells were stimulated with known PPARγ ligands (ciglitazone, pioglitazone, rosiglitazone, and troglitazone), along with the nonsynthetic ligand, 15-deoxy prostaglandin J2 (15d-PGJ2), all resulting in the phosphorylation of PPARγ1 (Fig. 1A). Pioglitazone was then used for both different durations and with varying concentrations to study the effect of PPARγ1 on phosphorylation. Concentrations as low as 50 nM yielded an altered phosphorylation pattern, and this phenomenon appeared to be PPARγ ligand-restricted because 100 μM PPARα-specific ligand WY-14643 did not have the same effect (Fig. 1B). Moreover, induced phosphorylation appeared within 30 min of stimulation, peaking at 1–2 h, without affecting the pool of total PPARγ (Fig. 1C). Notably, ligand-dependent phosphorylation was confirmed through the use of a ligand-binding inhibitor, GW9662, which efficiently prevented the phosphorylation of the receptor (Fig. 1D).

Synthetic Ligands Mimic Bacterial Induction of Phosphorylation, Which Correlates to Increased DNA Binding of PPARγ1.

Having observed ligand-induced phosphorylation of PPARγ1, we next performed EMSAs by using a PPARγ-specific probe on nuclear extracts from ligand-treated HT-29 cells. Importantly, increasing the pool of phospho-PPARγ1 protein resulted in elevated DNA binding, which in turn was prevented by the PPARγ antagonist, GW9662 (GW+Pio; Fig. 2A). Furthermore, a monoclonal antibody recognizing Ser-84 phosphorylation of PPARγ1 supershifted the majority of complex I, whereas the total PPARγ antibodies depleted both DNA-binding complexes (Fig. 2B). Hence, changes seen in DNA binding would primarily represent changes due to phosphorylation of PPARγ1. Oligonucleotide pulldown using a biotinylated DR-1 element from the CYP450 gene was used to monitor the phospho-PPARγ levels bound to DNA. In Fig. 2C, the levels of phosphorylated PPARγ binding to DNA increased in extracts from cells cocultured with rosiglitazone. EMSA also was performed on nuclear extracts from HT-29s cocultured with different isolates of Enterococcus faecalis, which are early colonizers of the human gut. All four isolates (EC1, EC3, EC15, and EC16) seemed to affect DNA binding of PPARγ1 (Fig. 2D). We decided to focus on one of the isolates, EC16, and established its rapid but transient effect on receptor DNA binding (data not shown). We proceeded to show that phospho-PPARγ1 is a component of the altered DNA-binding complex in EC16-stimulated cells through supershift assay (Fig. 2E). The same increase as with ligand stimuli in DNA binding can be seen after EC16 stimulation (Fig. 2F). Here we also show the presence of the PPARγ-binding partner, RXR, confirming proper DNA complex formation, along with the nonspecific DNA-binding protein, Ku70, as a loading control.

Increased Phospho-PPARγ1 Correlates to Increased Levels of Adipose Differentiation-Related Protein (ADRP), Fasting-Induced Adipose Factor (FIAF), and IL-10 Gene Expression.

Having demonstrated that resident flora from newborn children as well as synthetic ligands regulate the DNA binding of PPARγ1 in a phosphorylation-dependent manner, we next evaluated whether this phenomenon would correlate to the activation of downstream target genes by using SYBR Green-based real-time PCR. Time points of gene expression evaluation were set to 6 h to account for early transcriptional events. The effects on target gene expression were assessed, where the target gene, ADRP (or adipophilin), along with yet another PPARγ target gene, FIAF (or angiopoietin-like protein 4) (19, 20), were strongly induced by EC16 as well as a PPARγ-specific ligand, rosiglitazone (Fig. 3 A and B). In addition to ADRP and FIAF, we analyzed the effect of EC16 in IL-10, a cytokine essential for the maintenance of gut homeostasis. Here we show that IL-10 expression by HT-29 cells is increased when cocultured with both EC16 (Fig. 3C Left) and specific ligand (Fig. 3C Right). Furthermore, a decline in PPARγ protein levels, as achieved by specific siRNA knockdown experiments, indicated that EC16-induced FIAF expression is indeed dependent on PPARγ (Fig. 3D). The EC16-induced FIAF expression also was reviewed when testing the influence of the three different kinase inhibitors, PD98059, LY294002, and SB203580. This shows that the induction can be inhibited by LY294002 and partially by SB203580, whereas PD98059 seems unable to influence expression (Fig. 3E). HT-29s treated with EC16, together with either LY294002 or SBB203580, also show decreased levels of phospho-PPARγ1 (Fig. 3F), indicating that the PI3-kinase and, to some extent, the p38 MAPK pathways are instrumental in directing phosphorylation and, thus, activity.

Proteasomal Inhibition Sustains Phospho-PPARγ1 Levels and Prolongs DNA Binding of the Receptor Complex.

To establish whether the pool of phospho-PPARγ1 was influenced by degradation, HT-29 cells were subjected to the calpain inhibitor, ALLN, the 20S proteasome inhibitor, lactacystin, and a modifier of the proteasomal catalytic subunits, epoxomicin. All treatments led to the accumulation of the phosphorylated form of PPARγ1, differing only in the longevity of degradation inhibition (Fig. 4A). The amount of phosphorylated protein was elevated under the influence of MG-132 and even further so when costimulated with pioglitazone, which in turn could be inhibited by GW9662 (Fig. 4B). The disappearance of the phospho-PPARγ1-containing complex I in EMSAs at later time points (Fig. 2A) due to transient ligand effects can be impeded by MG-132 (Fig. 4C). PPARγ target gene transcription was subsequently assessed. Interestingly, the inhibition of the proteasome using MG-132 is enough to increase transcription; when costimulated with ligand, the response is synergistic. This synergism seen on ADRP expression as a result of MG-132 and ligand costimulation is effectively inhibited by GW9662, thus abrogating any ligand-specific effects (Fig. 4D). To further verify these findings, we performed similar stimulations on primary mouse colonocytes where EC16 and MG-132 both increase levels of phospho-PPARγ1 (Fig. 4E).

Cell Culture and Reagents.

The human colorectal adenocarcinoma cell line HT-29 (HTB-38; American Type Culture Collection) was grown and maintained according to the supplier's recommendations. Colons from male 10-week-old NMRI mice were scraped for epithelial cells and cultured in DMEM (Gibco).

All agonists and antagonists were dissolved in DMSO. Pioglitazone (5 μM), rosiglitazone (5 μM), troglitazone (5 μM), ciglitazone (5 μM), 15d-PGJ2 (10 μM), and GW-9662 (10 μM) were all acquired from Cayman Chemical. Cells were always pretreated with GW9662 for 1 h before further stimulation to avoid competition between compounds. The proteasome inhibitors MG-132 (10 μM), ALLN (25 μM), epoxomicin (100 nM), and lactacystin (5 μM) were all purchased from Calbiochem. The kinase inhibitors PD98059 (25 μM) and LY294002 (10 μM) were purchased from Calbiochem, while SB203580 (10 μM) came from Sigma–Aldrich. The PPARα ligand, WY-14643 (100 μM), was procured from Biomol. Antibodies for RXRα (sc-553), PPARγ (sc-7273 and sc-7196), and Actin (sc-1616) were obtained from Santa Cruz Biotechnology, and the anti-phospho-PPARγ-specific antibody, MAB-3632, was purchased from Chemicon.

The E. faecalis used here were isolates obtained from 1-month-old (EC1) and 3-day-old (EC3, EC15, and EC16) infants. Isolates were always added to cells at a concentration of 10⁶.

EMSA.

DNA binding was assayed by using 5 μg of nuclear extracts in binding buffer consisting of 25 mM Hepes (pH 7.9), 70 mM KCl, 10% glycerol, 5 mM DTT, and 1 μg of polydIdC (Amersham) in the presence of 50,000 cpm of a radiolabeled oligonucleotide probe. The probe chosen was an element of high PPARγ-specificity and moderate affinity to ensure low involvement of additional PPARs (43). The element used was the ARE6 DR1 site of the mouse adipocyte lipid-binding protein, aP2, of the following sequence: 5′-TCTCTCTGGGTGAAATGTGC-3′, 5′-AGAGGCACATTTCACCCAGAGAGA-3′. Supershift assays were carried out by initial incubation of the extracts in binding buffer, together with 500 ng of the antibody of interest, as well as suitable controls, for 25 min at room temperature. Thereafter, polydIdC and probe were added to extracts and incubated for 30 min before gel electrophoresis.

Oligonucleotide Pulldown Assay.

To trap DNA-binding complexes for further investigation, we used an oligonucleotide pulldown assay. Custom biotinylated oligonucleotides of a high-affinity PPRE in the CYP4A6/P450 IV family [CYP4A6(Z)] were designed: 5′-Biotin-CTGAACTAGGGCAAAGTTCACAGT-3′, 5′-biotin-ACTGTGAACTTTGCCCTAGTTCAG-3′. These were coupled (1 μg per sample) to streptavidin-coated Sepharose beads (Amersham Biosciences) and used to pull down DNA-binding PPARγ-RXR heterodimers from 80 μg of nuclear extracts in a binding buffer consisting of 12% glycerol, 12 mM Hepes (pH 7.9), 4 mM Tris (pH 7.9), 150 mM KCl, 1 mM EDTA, 1 mM DTT, 10 μg poly (dIdC)/300 μg protein, 1.2 mM PMSF, and 1x Complete (Roche).

Real-Time PCR.

Total RNA was extracted by using an RNeasy mini-kit (Qiagen), and cDNA synthesis was performed by using SuperScript II (Invitrogen), both according to the manufacturer's instructions. We have designed and tested a selection of annotated and potential target genes for PPARγ by using SYBR Green (Applied Biosystems)-based semiquantitative real-time PCR. Primers were designed and tested according to Applied Biosystems recommendations: ADRP forward, 5′-CTGTTCACCTGATTGAATTTGC-3′, reverse, 5′-AGAGCTTATCCTGAGCATCCTG-3′; β-actin forward, 5′-CCTGGCACCCAGCACAAT-3′, reverse, 5′-gccgatccacacggagtact-3′; FIAF forward, 5′-AAAGAGGCTGCCCGAGAT-3′, reverse, 5′-TCTCCCCAACCTGGAACA-3′; IL-10 forward, 5′-TCCCTGTGAAAACAAGAGCA-3; reverse, 5-TGTCAAACTCACTCATGGCTTT-3′. The sample setups always included biological duplicates and experimental triplicates. Data are presented as fold change in the relative gene expression.

RNA Interference.

The siRNA for PPARγ and scrambled control was purchased from Dharmacon, along with the transfection reagent suitable for the HT-29 cell line (Dharmafect 4). Transfection efficiency was always confirmed through either Western blot or real-time PCR and was carried out at least 24 h before any further stimulation of cells.

Western Blot Analysis.

Cells treated according to figure specifications were harvested and lysed. Nuclear extracts were prepared according the nuclear extracts protocol of Schreiber (44), whereas whole-cell extracts were made with Schindler lysis buffer [50 mM Tris (pH 8), 0.1 mM EDTA, 0.5% Nonidet P-40, 10% glycerol, and 150 mM NaCl]. To prevent dephosphorylation in prepared extracts, 10 nM okadaic acid, 5 mM sodium fluoride, and 400 μM sodium vanadate were added, along with 1× Complete (Roche) and 1 mM PMSF. Immunodetection with an appropriate secondary peroxidase-conjugated antibody (DAKO) was followed by chemiluminescence (ECL; Amersham). Western blots were always sequentially probed with phospho-PPARγ, total PPARγ.

Quantification of protein bands was performed by using the Quantity One analysis software of the Discovery Series (Bio-Rad).