- Email: [email protected]
SMRT recruitment by PPARγ is mediated by specific residues located in its carboxy-terminal interacting domain
Molecular and Cellular Endocrinology 267 (2007) 138–143
Erratum
SMRT recruitment by PPAR␥ is mediated by specific residues located in its carboxy-terminal interacting domain Maria M. Sutanto, Melissa S. Symons, Ronald N. Cohen ∗ Section of Endocrinology, Department of Medicine, The University of Chicago, 5841 S. Maryland Avenue, MC 1027, Chicago, IL 60637, United States
Abstract The silencing mediator of retinoid and thyroid hormone receptors (SMRT) has been shown to play an important role in adipogenesis and PPAR␥ transcriptional activity. SMRT contains two interacting domains that mediate interactions with nuclear receptors. Interestingly, SMRT is recruited to PPAR␥ via its C-terminal interacting domain, and mutation of the proximal interacting domain does not interfere with recruitment via PPAR␥. To understand how the distal interacting domain mediates recruitment by PPAR␥, we have now mutated residues in this domain to the corresponding amino acids found in the proximal domain. We show that specific residues in this distal domain are vital for interactions with PPAR␥, but not for a related receptor, RAR␣. Furthermore, naturally occurring SMRT isoforms that differ in interacting domain sequences have different effects on PPAR␥ as opposed to RAR␣ recruitment. These data suggest that PPAR␥ and RAR␣ interact with SMRT via distinct mechanisms. These differences will be important as ligands are designed that lead to specific patterns of nuclear receptor recruitment of corepressors. Published by Elsevier Ireland Ltd. Keywords: Corepressors; Peroxisome proliferator-activated receptor; Adipogenesis; Silencing mediator of retinoid and thyroid hormone receptors; Transcriptional regulation
1. Introduction The peroxisome proliferator-activated receptor ␥ (PPAR␥) is a member of the nuclear receptor superfamily that plays an important role in adipogenesis and insulin sensitivity (Lehrke and Lazar, 2005). The thiazolidinediones (TZDs) are a class of medications used in the treatment of type 2 diabetes mellitus that serve as exogenous ligands for PPAR␥ and lead to an increase in coactivator recruitment to the receptor. Nuclear receptor coactivators, via effects on histone acetylation and other processes, lead to an increase in gene transcription. Thus, by stimulating PPAR␥ transcriptional activity, TZDs increase expression of PPAR␥ target genes. Another class of cofactors, nuclear receptor corepressors (CoRs), have been shown to down-regulate nuclear receptor-mediated gene transcription. The two major CoRs are the silencing mediator of retinoid and thyroid hormone receptors (SMRT) (Chen and Evans, 1995) and the nuclear receptor corepressor (NCoR) (Horlein et al., 1995). Recent work suggests that SMRT and NCoR play important roles in PPAR␥ action. Data from our group and others have shown that SMRT and
∗
DOI of original article:10.1016/j.mce.2006.08.004. Corresponding author. Tel.: +1 773 834 1012; fax: +1 773 834 0486. E-mail address: [email protected] (R.N. Cohen).
0303-7207/$ – see front matter. Published by Elsevier Ireland Ltd. doi:10.1016/j.mce.2006.10.015
NCoR repress PPAR␥-mediated gene transcription on a subset of target genes (Guan et al., 2005; Yu et al., 2005). By repressing PPAR␥ activity, SMRT and NCoR modulate adipogenesis (Yu et al., 2005). SMRT and NCoR share a similar overall structure (Privalsky, 2004). Each is composed of amino-terminal repressing domains and carboxy-terminal interacting domains (IDs) (Fig. 1A). These IDs (cross-hatched areas in Fig. 1A) mediate interactions with nuclear receptors. SMRT contains two interacting domains, a distal (S1) and a proximal (S2) domain. In contrast, NCoR contains three IDs. Each ID in SMRT and NCoR contains a unique sequence, called a CoRNR box motif (each represented by a gray box in Fig. 1A), that is critical for nuclear receptor–CoR interactions. CoRNR box motifs contains the consensus sequence L–x–x–I/H–I–x–x–x–I/L (also reported as I/L–x–x–I/V–I) (Hu and Lazar, 1999; Nagy et al., 1999; Perissi et al., 1999). The fact that each corepressor ID contains a unique CoRNR box sequence (see below and Fig. 1C) suggests there may be specificity in terms of corepressor recruitment by nuclear receptors. Early work suggested that TR preferred to recruit NCoR whereas RAR␣ preferred SMRT (Cohen et al., 2000). Recent work by Goodson et al. (2005) has suggested that corepressor preference is more complex than previously thought because SMRT consists of two isoforms that differ in their interacting domains. These two isoforms have been termed SMRT␣ and SMRT;
M.M. Sutanto et al. / Molecular and Cellular Endocrinology 267 (2007) 138–143
139
interestingly, SMRT binds TR weakly, but SMRT␣ binds TR more strongly (Goodson et al., 2005). Our prior data suggests that PPAR␥ recruits SMRT specifically through its distal S1 domain, suggesting there is clear ID preference for PPAR␥ (Yu et al., 2005). In the current study, we have studied the mechanisms for PPAR␥–Sl interactions. We mutated residues in SMRT ID1 to resemble those of SMRT ID2. Interestingly, some but not all of these alterations block SMRT recruitment by PPAR␥. Thus, PPAR␥ depends on certain crucial S1 CoRNR box sequences to recruit SMRT. In contrast, mutation of these residues in SMRT ID1 does not block RAR␣–SMRT interactions. These data confirm that SMRT interacts differently with PPAR␥ as opposed to RAR␣, and suggest there may be mechanisms to modulate corepressor recruitment by distinct nuclear receptors. Finally, and in contrast to TR, we show that PPAR␥ interacts preferentially with SMRT over SMRT␣, suggesting another potential mechanism for influencing corepressor recruitment by nuclear receptors. 2. Materials and methods 2.1. Plasmids Upstream activating sequence (UAS)-Luc contains five copies of a Gal4binding site upstream of luciferase. Gal4-SMRT and Gal4-NCoR contain the IDs of SMRT and NCoR, respectively, downstream of Gal4 and have been previously described (Cohen et al., 2000). The Gal4-SMRT mutant constructs were made in Gal4-SMRT. The mutant Gal4-SMRT constructs were created by site-directed mutagenesis using the Quick-Change Mutagenesis Kit (Stratagene). Gal4-SMRT␣ was made from the corresponding region of SMRT␣, which was a generous gift of Martin Privalsky (University of California). The Gal4PPAR␥, Gal4-RAR␣, and Gal4-RXR␣ constructs contain the ligand-binding domains of the nuclear receptors downstream of Gal4. The constructs for VP16PPAR␥2 (Yu et al., 2005) and VP16-RAR␣ (Cohen et al., 2000) contains the full-length nuclear receptors downstream of the VP16 activating domain in the vectors pVP16 and AASV-VP16, respectively. PPRE-Luc has been described previously (Yu et al., 2005).
2.2. Cell culture and transfection 3T3-L1 and CV-1 cells were grown in DMEM supplemented with antibiotics, glutamine, and 10% calf serum or fetal calf serum, respectively. Cells were transfected in 6-well plates using lipofectamine (Invitrogen) in serum-free media, as per manufacturers’ instructions. Three to four hours after transfection, media was changed to fresh serum-free media, and cells extracts were processed the following day for luciferase activity. All transfections were performed in triplicate and repeated at least three times. Fig. 1. Mutation of SMRT ID1 blocks recruitment by PPAR␥: (A) schematic diagram of SMRT and NCoR. SMRT contains two C-terminal interacting domains (ID1 and ID2); NCoR contains three such domains (ID1–ID3). IDs are represented by cross-hatched boxes. Each ID contains a unique CoRNR box sequence (gray boxes see (C) for sequence details); (B) 3T3-L1 cells were transfected with 300 ng UAS-Luc reporter along with 300 ng Gal4-PPAR␥ or Gal4-RXR␣, and 300 ng pSG5-SMRT (or empty pSG5). Cell extracts were obtained 24 h after transfection and assayed for luciferase activity. Data are expressed as relative luciferase activity, where 1.0 represents luciferase activity in the absence of cotransfected pSG5-SMRT; (C) amino acid residues of SMRT and NCoR CoRNR box motifs, based on the consensus sequence of Picard et al.; (D) 3T3L1 cells were transfected with 300 ng UAS-luciferase reporter plasmid along with 300 ng Gal4-SMRT or Gal4-SMRT mutant construct, and 300 ng VP16PPAR␥2 or empty VP16. Data are presented as fold luciferase activity, i.e., the induction of luciferase activity in the presence of the Gal4 construct with VP16-PPAR␥2 compared to that with the empty VP16 vector.
2.3. GST interaction assays GST-S1 and GST-S2 have been previously described (Cohen et al., 2000). GST fusion proteins were expressed in DH5␣ E. coli by induction with 0.1 mM isopropylthio--d-galactosidase at 30 ◦ C. Proteins were isolated after disruption with French Press and purified on Sepharose beads with extensive washing. PPAR␥2 was in vitro translated in reticulocyte lysate (Promega) using T7 polymerase. A 5 ml of 35 S-labeled PPAR␥2 was incubated with GST-corepressor constructs. After extensive washing, bound proteins were eluted by boiling in loading buffer and analyzed by SDS-PAGE.
2.4. RT-PCR mRNA from 3T3-L1 cells was obtained using Micro Fast-Track mRNA Isolation Kit (Invitrogen). RT-PCR for the SMRT isoforms was performed using the
140
M.M. Sutanto et al. / Molecular and Cellular Endocrinology 267 (2007) 138–143
Access RT-PCR system (Promega). The primers used to show relative expression of SMRT␣ and SMRT were the same as previously published by Goodson et al. (2005).
3. Results 3.1. SMRT recruitment to PPARγ-RXR heterodimers is dependent on distinct residues within the SMRT ID1 domain We have previously shown that both SMRT and NCoR downregulate PPAR␥-mediated gene transcription. PPAR␥ binds DNA as a PPAR␥-RXR heterodimer. To determine if SMRT effects are primarily due to the PPAR␥ or RXR components of the heterodimer, we used the Gal4-system. In this way, Gal4PPAR␥ or Gal4-RXR␣ constructs could be tested individually in 3T3-L1 cells. 3T3-L1 fibroblast cells were chosen in these experiments because they have been shown to differentiate into adipocytes under hormonal stimulation. As shown in Fig. 1B, overexpression of SMRT represses Gal4-PPAR␥ transcriptional activity, but is less effective with Gal4-RXR. These data suggest that corepressors do not mediate their effects on PPAR␥ primarily via interactions with the RXR heterodimeric partner. Each corepressor ID contains a unique CoRNR box motif, which is characterized by the sequence L–x–x–I/H–I–x–x–x–I/L (Fig. 1C) (note that in certain papers the CoRNR box motif is represented by the sequence I/L–x–x–I/V–I; since the former version includes the extended helical motif, it will be used here) (Hu and Lazar, 1999; Nagy et al., 1999; Perissi et al., 1999). This is particularly true for the domains SMRT ID1 (S1), SMRT ID2 (S2), NCoR ID1 (N1), and NCoR ID2 (N2). The most proximal NCoR ID, ID3 (N3), contains a variant on this sequence and has no homologous region in SMRT; thus, this manuscript will focus on the other IDs. We have previously shown that PPAR␥ recruits SMRT in 3T3-L1 cells primarily via its S1 domain (Yu et al., 2005). Similarly, the homologous domain of NCoR (N1) is required for full interactions of NCoR and PPAR␥ (data not shown). We have focused on SMRT in the current manuscript since the S1 domain can be altered to resemble S2, and there are only two such domains. Importantly, many CoRNR box amino acid residues are similar in S1 and N1. For example, the residue at the +2 location of the CoRNR box sequence is either a glutamic acid (E) in S1 and N1, or an alanine (A) in S2 and N2. The amino acids at positions +4 and +9 are also conserved. The alanine (A) at position +8 of the S1 and N1 CoRNR box is relatively conserved, in that the homologous residue is a valine (V) in S2 and an isoleucine (I) in N2. These striking similarities between the distal CoRNR box sequences suggest that there are specific amino acid residues in S1 and N1 that are vital for optimal PPAR␥ interactions. Therefore, we mutated amino acids in SMRT ID1 to the corresponding residues in SMRT ID2 to determine which residues are important for full interactions with PPAR␥. To define these interactions between PPAR␥ and mutant SMRT constructs, we performed mammalian two-hybrid experiments in 3T3-L1 cells. 3T3-L1 cells were chosen because they are a well-recognized model for adipogenesis, and are capable of differentiating into adipocytes under the correct hormonal stim-
Fig. 2. Mutation of SMRT blocks SMRT repression of PPAR␥ transcriptional activity: (A) 35 S labeled PPAR␥2 was incubated with SMRT ID constructs, GST-ID1, GST-ID1mut, and GST-ID2. After extensive washing, bound proteins were evaluated by SDS-PAGE upper panel. Mutation of the +4 isoleucine in the ID1 CoRNR box significantly decreases interactions with PPAR␥. The lower panel shows that GST constructs were made in similar amounts, though slightly more GST-ID2 was made than GST-ID1 or GST-ID1mut; (B) each well of 3T3L1 cells was cotransfected with 300 ng of a PPRE-luciferase construct, 300 mg pSG5-PPAR␥2, and 300 ng pSG5, pSG5-SMRT, or pSG5-SMRTmut. Data are expressed as relative luciferase activity, where 1.0 is defined as luciferase activity in the presence of PPRE-Luc, pSG5-PPAR␥2 and empty pSG5 without SMRT or SMRTmut.
ulation. As shown in Fig. 1D, changing the isoleucine at position +4 of SMRT ID1 to histidine completely blocked PPAR␥–SMRT interactions in 3T3-L1 cells. In addition, mutation of glutamic acid to alanine also dramatically decreased interactions, though not to the degree of the isoleucine mutation. In contrast, mutation of alanine to valine did not decrease interactions at all, suggesting that the specific amino acid in this position is not as vital to interactions with PPAR␥, or that either alanine or valine is sufficient for optimal interactions. Next, the isoleucine-to-histidine mutation was inserted into GST-ID1 to determine if this mutation decreases direct recruitment of the isolated interacting domains to PPAR␥. The GST-ID1 and GST-ID2 have been previously described (Cohen et al., 2000). These constructs were incubated with in vitro translated 35 S-labeled PPAR␥2, extensively washed, and analyzed by SDS-PAGE. As shown in Fig. 2A (upper panel), GST-ID1 interacts strongly with PPAR␥, but GST-ID2 interacts weakly if at all with PPAR␥. Interestingly, mutation of the isoleucine at position +4 of SMRT ID1 to the corresponding histidine of ID2 dramatically decreases interactions of ID1 with PPAR␥. The lower panel of Fig. 2A shows that the GST-ID1 and GST-ID1mut constructs were made in similar amounts; although slightly more GST-ID2 was made than GST-ID1, GST-ID2 still did not interact with PPAR␥2. These data confirm that the isoleucine at +4 is vital for SMRT recruitment by PPAR␥ and suggest that the structures of ID1 and ID2 differ in important ways to determine optimal recruitment by nuclear receptors. To determine if the mutation would block SMRT-mediated repression of full-length PPAR␥ transcriptional activity, the
M.M. Sutanto et al. / Molecular and Cellular Endocrinology 267 (2007) 138–143
isoleucine-to-histidine mutation was placed into pSG5-SMRT. These constructs were cotransfected in 3T3-L1 cells with pSG5PPAR␥2 and a PPRE-luciferase reporter. As shown in Fig. 2B, overexpression of pSG5-SMRT decreased PPAR␥-mediated luciferase activity. In contrast, this action was blocked after mutation of the isoleucine of the SMRT ID1 CoRNR box to the corresponding histidine of ID2. These data suggest that func-
141
tional interactions of SMRT with PPAR␥2 are dependent on the ID1 CoRNR box sequence. 3.2. Mutation of the SMRT ID1 isoleucine at position +4 blocks SMRT repression of PPARγ transcriptional activity, but does not block SMRT effects on RARα To determine if the same SMRT ID1 mutation would block interactions with RAR␣, a mammalian two-hybrid assay with VP16-RAR␣ was performed in 3T3-L1 cells. As shown in Fig. 3A, Gal4-SMRT but not Gal4-SMRT I–>H again interacted with VP16-PPAR␥2. In contrast, both Gal4-SMRT and Gal4-SMRT I–>H interacted well with VP16-RAR␣. To compare the functional effects of the SMRT mutation on PPAR␥ and RAR␣ activity, pSG5-SMRT (or pSG5 alone) was cotransfected with Gal4-empty vector or Gal4-PPAR␥ in 3T3-L1 cells. As shown in Fig. 3B, overexpression of SMRT decreased Gal4-PPAR␥ activity. In contrast, overexpression of the SMRT I–>H mutation did not lead to repression. As a control, both constructs did not repress an empty Gal4 construct. We next wanted to determine if the SMRT constructs would lead to repression of RAR. Interestingly, Gal4-RAR␣ exhibited very low levels of transcriptional activity in 3T3-L1 cells, and overexpression of SMRT or the SMRT mutant did not lead an enhancement of repression in these cells. Therefore, we performed the RAR experiments in CV-1 cells, since SMRT has
Fig. 3. A SMRT mutation is specific for interactions with PPAR␥: (A) 3T3L1 cells were transfected with 300 ng UAS-luciferase; 300 ng Gal4-SMRT or Gal4-SMRT (I–>H); 300 ng VP16-PPAR␥2 or VP16-RAR␣ (or the appropriate empty VP16 vector). Data are expressed as fold luciferase activity, i.e., the induction of luciferase activity with VP16-PPAR␥2 (or VP16-RAR␣) compared to that of the empty VP16 vector; (B) 3T3-L1 cells were transfected with 300 ng UAS-luciferase; 300 ng Gal4-PPAR␥ (or empty Gal4); 300 ng empty pSG5, pSG5-SMRT, or pSG5-SMRT (I–>H). Data are expressed as relative luciferase activity, where 1.0 is defined as luciferase activity in the presence of empty pSG5 and empty Gal4 (upper panel) or empty pSG5 and Gal4-PPAR␥ (lower panel); (C) CV-1 cells were transfected as in B, except Gal4-RAR␣ was used at a dose of 100 ng due to differences in transfections between CV-1 and 3T3-L1 cells.
Fig. 4. PPAR␥ prefers to interact with the SMRT isoform: (A) amino acid sequence of the portion of S1 containing differences between SMRT and SMRT␣. Both IDs contain the appropriate S1 CoRNR box as shown. After the CoRNR box, SMRT␣ contains 47 extra amino acids (shown in bold) as compared with SMRT; (B) RT-PCR or 3T3-L1 cells showing the presence of both SMRT and SMRT␣; (C) 3T3-L1 cells were transfected with 300 ng UAS-luciferase; 100 ng Gal4-SMRTt or Gal4-SMRT␣; VP16-PPAR␥2 or VP16-RAR␣ (or the corresponding empty VP16 vector). Data are expressed as fold luciferase activity in the presence of the VP16-PPAR␥2 (pr VP16-RAR␣) construct over that of the empty VP16 vector.
142
M.M. Sutanto et al. / Molecular and Cellular Endocrinology 267 (2007) 138–143
previously been shown to enhance RAR␣-mediated repression in this cell line by other groups (Hauksdottir et al., 2003). In CV1 cells, cotransfection of both pSG5-SMRT and pSG5-SMRT I–>H led to enhanced repression of RAR␣ (Fig. 3C). Thus, SMRT interacts with RAR␣ differently than it does with PPAR␥. 3.3. PPARγ preferentially interacts with the SMRT isoform Recently, Goodson et al. (2005) has shown that SMRT consists of two isoforms, which differ in their sequences just distal to the SMRT ID CoRNR box (Fig. 4A). Our previous work above used SMRT constructs (Cohen et al., 2001; Yu et al., 2005). Thus, we were interested in determining if PPAR␥ could recruit SMRT␣, and how this interaction compared to that with SMRT. Using RT-PCR, we first determined which SMRT isoforms are present in 3T3-L1 cells. In fact, both SMRT␣ and SMRT are present in 3T3-L1 cells, though SMRT appears to be more abundant (Fig. 4B). Next, we performed a mammalian two-hybrid assay to determine whether PPAR␥ could recruit SMRT␣ to a similar degree as SMRT. The portion of SMRT␣ (a generous gift of Martin Privalsky) that was homologous to the SMRT region used previously in Gal4-SMRT was placed downstream of Gal4 to make Gal4-SMRT␣. Gal4-SMRT␣ and Gal4-SMRT were cotransfected with VP16-PPAR␥2 (or VP16 alone). As shown in Fig. 4C, PPAR␥ interacted well with both SMRT isoforms; however, its interactions with SMRT were stronger. In contrast, RAR␣ interacted equally well with both SMRT isoforms in 3T3-L1 cells, similar to previous data in other cell types (Goodson et al., 2005). These data suggest that the PPAR␥ specificity for SMRT ID1 extends to naturally occurring SMRT isoforms that differ in their ID1 sequences.
S1 and S2 are important for PPAR␥ interactions. We found that some, but not all, such residues that are conserved in S1 and N1, but are different in S2, are vital for interactions with PPAR␥. In particular, the isoleucine at position +4 of the S1 CoRNR box sequence was crucial for strong interactions of SMRT with PPAR␥. Mutation of the isoleucine to the corresponding histidine found in S2 blocked SMRT recruitment by PPAR␥, even though all other amino acids in S1 were maintained. When this mutation was introduced into SMRT, SMRT-mediated repression of PPAR␥ was also abolished. The SMRT corepressor is also a major determinant of RAR␣ action. RAR␣ is a nuclear receptor which (similar to PPAR␥) heterodimerizes with RXR. In addition, the SMRT corepressor is also a major determinant of RAR␣ transcriptional activity. RAR␣ binds vitamin A metabolites as ligand (Mangelsdorf et al., 1995). RAR␣ represses gene transcription in the absence of ligand, and this repression is mediated by SMRT and NCoR (Cohen et al., 2000; Hauksdottir et al., 2003). Previous work has shown that RAR␣ interacts better with SMRT than with NCoR, and some of this preference may be mediated by the S2 domain (Cohen et al., 2001). Interestingly, mutation of S1 CoRNR box isoleucine at position +4 to histidine did not decrease RAR␣–SMRT interactions, and the mutant SMRT was still able to repress RAR␣ transcriptional activity. Thus, RAR␣ and PPAR␥ interact differently with SMRT, even though both interactions occur via SMRT IDs. The crystal structure of SMRT complexed to PPAR␥ has not yet been identified. However, the structure of the related receptor PPAR␣ with a portion of SMRT ID1 in the presence of an antagonist has been determined (Xu et al., 2002). Xu et al. showed that the SMRT ID1 CoRNR box sequence forms a three-turn alpha helix that fits into a groove formed by PPAR␣ helices 3, 3 , 4, and 5. This allows the SMRT ID1 residues L1, I5, and
4. Discussion PPAR␥ has been shown to regulate adipogenesis and modulate insulin sensitivity. The corepressor SMRT binds PPAR␥ and represses its transcriptional activity (Cohen, 2006; Guan et al., 2005; Yu et al., 2005). The binding of SMRT to PPAR␥ is regulated at a number of levels, for example, the degree of ligand binding (Yu et al., 2005) and the specific nature of the underlying DNA response element (Guan et al., 2005). The recruitment of SMRT has been shown to affect the ability of TZDs to stimulate adipogenesis (Yu et al., 2005), and it is likely that the interactions between SMRT and PPAR␥ will be important in understanding other PPAR␥-mediated effects such as the regulation of insulin sensitivity. Thus, it is important to understand how SMRT is recruited to PPAR␥. We have focused on the specific amino acid residues in SMRT ID1 (S1). We have previously shown that SMRT binding to PPAR␥ specifically depends on this domain. Furthermore, the CoRNR box sequence of this domain has a number of interesting similarities to that of NCoR ID1, which primarily mediates interactions between NCoR and PPAR␥. Prior data suggests that the other SMRT ID, S2, does not play an important role in SMRT–PPAR␥ interactions (Yu et al., 2005). Thus, we examined the amino acid residues of the S1 CoRNR box sequence to see which residues that differ between
Fig. 5. Schematic diagram of SMRT ID1 CoRNR box complexed to the PPAR␣ ligand binding domain. The structure is derived from PubMed structure PDB 1KKQ (mmdb 18993). The amino acid residues of SMRT ID1 that are included in the structure are represented by spheres (with the CoNR box residues in yellow, and other residues in purple). The core residues (L1, I5, and L9), as well as the mutated residue I4 are indicated with arrows. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)
M.M. Sutanto et al. / Molecular and Cellular Endocrinology 267 (2007) 138–143
L9 to be aligned. In contrast, the residues I4, A8, E2, and R6 are oriented at 90◦ angles to this interface (Fig. 5); however, I4 makes additional hydrophobic contacts along the binding groove (Xu et al., 2002), and is thus in position to allow for differential interactions with nuclear receptors. Goodson et al. (2005) have recently described two SMRT isoforms that differ in S1 sequences just distal to the S1 CoRNR box. We hypothesized that the recruitment of SMRT by PPAR␥ would be affected by differences between these two isoforms. In fact, PPAR␥ recruited SMRT more strongly than SMRT␣ in a two-hybrid experiment. SMRT is the isoform that lacks 47 amino acids present in SMRT␣. This is distinct from RAR␣, which recruited both SMRT isoforms equally well. This contrasts with the TR, which has been shown to bind SMRT␣ better than SMRT (Goodson et al., 2005). Our data suggests that 3T3L1 cells contain both SMRT isoforms. It is possible that one or the other SMRT isoform might be preferentially expressed under certain circumstances, which might allow differing transcriptional effects in the cell. Thus, it might be possible to alter adipogenesis via changing the proportional amounts of SMRT versus SMRT␣ in the cell. Although our data are limited to PPAR␥–SMRT isoform interaction assays, we anticipate that these differences in interactions will be reflected in the functional ability of the SMRT isoforms to repress PPAR␥ transcription. We have not yet been able to perform those experiments, since our full-length SMRT constructs are derived from different species, and are thus beyond the scope of the current manuscript. PPAR␥ has been shown to have play multiple roles in vivo. Its best characterized function is as the central regulators of adipogenesis (Lehrke and Lazar, 2005). More recently, PPAR␥ has also been implicated in other tissues, such as the immune system (Pascual et al., 2005), muscle (Hevener et al., 2003; Norris et al., 2003), liver (Matsusue et al., 2003), and the gastrointestinal tract (Auwerx, 2002). TZDs serve as ligands for PPAR␥. Selective PPAR␥ modulators (SPPARMs) may be developed with differing profiles of transcriptional activation (Auwerx, 2002). SPPARMs enhance coactivator recruitment in certain tissues or on certain promoters, and corepressor recruitment on others. Thus, it is important to understand those factors that lead to corepressor recruitment by PPAR␥. A further understanding of how PPAR␥ recruits SMRT, and how PPAR␥ differs from other nuclear receptors in this fashion, will eventually allow us to define ways to modulate SMRT effects on adipogenesis and insulin sensitivity. Acknowledgements This work was supported by R21 DK072397. We gratefully acknowledge Martin Privalsky (SMRT␣) and J. Don Chen and Ronald Evans (SMRT) for plasmids. We thank Katie Markan for technical assistance. References Auwerx, J., 2002. Nuclear receptors. I. PPAR gamma in the gastrointestinal tract: gain or pain? Am. J. Physiol. Gastrointest. Liver Physiol. 282, G581–G585.
143
Chen, J.D., Evans, R.M., 1995. A transcriptional co-repressor that interacts with nuclear hormone receptors. Nature 377, 454–457. Cohen, R.N., 2006. Nuclear receptor corepressors and PPARgamma. Nucl. Recept. Signal 4, e003. Cohen, R.N., Brzostek, S., Kim, B., Chorev, M., Wondisford, F.E., Hollenberg, A.N., 2001. The specificity of interactions between nuclear hormone receptors and corepressors is mediated by distinct amino acid sequences within the interacting domains. Mol. Endocrinol. 15, 1049–1061. Cohen, R.N., Putney, A., Wondisford, F.E., Hollenberg, A.N., 2000. The nuclear corepressors recognize distinct nuclear receptor complexes. Mol. Endocrinol. 14, 900–914. Goodson, M.L., Jonas, B.A., Privalsky, M.L., 2005. Alternative mRNA splicing of SMRT creates functional diversity by generating corepressor isoforms with different affinities for different nuclear receptors. J. Biol. Chem. 280, 7493–7503. Guan, H.P., Ishizuka, T., Chui, P.C., Lehrke, M., Lazar, M.A., 2005. Corepressors selectively control the transcriptional activity of PPARgamma in adipocytes. Gene Dev. 19, 453–461. Hauksdottir, H., Farboud, B., Privalsky, M.L., 2003. Retinoic acid receptors beta and gamma do not repress, but instead activate target gene transcription in both the absence and presence of hormone ligand. Mol. Endocrinol. 17, 373–385. Hevener, A.L., He, W., Barak, Y., Le, J., Bandyopadhyay, G., Olson, P., Wilkes, J., Evans, R.M., Olefsky, J., 2003. Muscle-specific Pparg deletion causes insulin resistance. Nat. Med. 9, 1491–1497. Horlein, A.J., Naar, A.M., Heinzel, T., Torchia, J., Gloss, B., Kurokawa, R., Ryan, A., Kamei, Y., Soderstrom, M., Glass, C.K., Rosenfeld, M.G., 1995. Ligand-independent repression by the thyroid hormone receptor mediated by a nuclear receptor co-repressor. Nature 377, 397–404. Hu, X., Lazar, M.A., 1999. The CoRNR motif controls the recruitment of corepressors by nuclear hormone receptors. Nature 402, 93–96. Lehrke, M., Lazar, M.A., 2005. The many faces of PPARgamma. Cell 123, 993–999. Mangelsdorf, D.J., Thummel, C., Beato, M., Herrlich, P., Schutz, G., Umesono, K., Blumberg, B., Kastner, P., Mark, M., Chambon, P., Evans, R.M., 1995. The nuclear receptor superfamily: the second decade. Cell 83, 835– 839. Matsusue, K., Haluzik, M., Lambert, G., Yim, S.H., Gavrilova, O., Ward, J.M., Brewer Jr., B., Reitman, M.L., Gonzalez, F.J., 2003. Liver-specific disruption of PPARgamma in leptin-deficient mice improves fatty liver but aggravates diabetic phenotypes. J. Clin. Invest. 111, 737–747. Nagy, L., Kao, H.Y., Love, J.D., Li, C., Banayo, E., Gooch, J.T., Krishna, V., Chatterjee, K., Evans, R.M., Schwabe, J.W., 1999. Mechanism of corepressor binding and release from nuclear hormone receptors. Gene Dev. 13, 3209–3216. Norris, A.W., Chen, L., Fisher, S.J., Szanto, I., Ristow, M., Jozsi, A.C., Hirshman, M.F., Rosen, E.D., Goodyear, L.J., Gonzalez, F.J., et al., 2003. Musclespecific PPARgamma-deficient mice develop increased adiposity and insulin resistance but respond to thiazolidinediones. J. Clin. Invest. 112, 608–618. Pascual, G., Fong, A.L., Ogawa, S., Gamliel, A., Li, A.C., Perissi, V., Rose, D.W., Willson, T.M., Rosenfeld, M.G., Glass, C.K., 2005. A SUMOylationdependent pathway mediates transrepression of inflammatory response genes by PPAR-gamma. Nature 437, 759–763. Perissi, V., Staszewski, L.M., McInerney, E.M., Kurokawa, R., Krones, A., Rose, D.W., Lambert, M.H., Milburn, M.V., Glass, C.K., Rosenfeld, M.G., 1999. Molecular determinants of nuclear receptor–corepressor interaction. Gene Dev. 13, 3198–3208. Privalsky, M.L., 2004. The role of corepressors in transcriptional regulation by nuclear hormone receptors. Annu. Rev. Physiol. 66, 315–360. Xu, H.E., Stanley, T.B., Montana, V.G., Lambert, M.H., Shearer, B.G., Cobb, J.E., McKee, D.D., Galardi, C.M., Plunket, K.D., Nolte, R.T., et al., 2002. Structural basis for antagonist-mediated recruitment of nuclear co-repressors by PPARalpha. Nature 415, 813–817. Yu, C., Markan, K., Temple, K.A., Deplewski, D., Brady, M.J., Cohen, R.N., 2005. The nuclear receptor corepressors NCoR and SMRT decrease peroxisome proliferator-activated receptor gamma transcriptional activity and repress 3T3-L1 adipogenesis. J. Biol. Chem. 280, 13600– 13605.
Erratum
SMRT recruitment by PPAR␥ is mediated by specific residues located in its carboxy-terminal interacting domain Maria M. Sutanto, Melissa S. Symons, Ronald N. Cohen ∗ Section of Endocrinology, Department of Medicine, The University of Chicago, 5841 S. Maryland Avenue, MC 1027, Chicago, IL 60637, United States
Abstract The silencing mediator of retinoid and thyroid hormone receptors (SMRT) has been shown to play an important role in adipogenesis and PPAR␥ transcriptional activity. SMRT contains two interacting domains that mediate interactions with nuclear receptors. Interestingly, SMRT is recruited to PPAR␥ via its C-terminal interacting domain, and mutation of the proximal interacting domain does not interfere with recruitment via PPAR␥. To understand how the distal interacting domain mediates recruitment by PPAR␥, we have now mutated residues in this domain to the corresponding amino acids found in the proximal domain. We show that specific residues in this distal domain are vital for interactions with PPAR␥, but not for a related receptor, RAR␣. Furthermore, naturally occurring SMRT isoforms that differ in interacting domain sequences have different effects on PPAR␥ as opposed to RAR␣ recruitment. These data suggest that PPAR␥ and RAR␣ interact with SMRT via distinct mechanisms. These differences will be important as ligands are designed that lead to specific patterns of nuclear receptor recruitment of corepressors. Published by Elsevier Ireland Ltd. Keywords: Corepressors; Peroxisome proliferator-activated receptor; Adipogenesis; Silencing mediator of retinoid and thyroid hormone receptors; Transcriptional regulation
1. Introduction The peroxisome proliferator-activated receptor ␥ (PPAR␥) is a member of the nuclear receptor superfamily that plays an important role in adipogenesis and insulin sensitivity (Lehrke and Lazar, 2005). The thiazolidinediones (TZDs) are a class of medications used in the treatment of type 2 diabetes mellitus that serve as exogenous ligands for PPAR␥ and lead to an increase in coactivator recruitment to the receptor. Nuclear receptor coactivators, via effects on histone acetylation and other processes, lead to an increase in gene transcription. Thus, by stimulating PPAR␥ transcriptional activity, TZDs increase expression of PPAR␥ target genes. Another class of cofactors, nuclear receptor corepressors (CoRs), have been shown to down-regulate nuclear receptor-mediated gene transcription. The two major CoRs are the silencing mediator of retinoid and thyroid hormone receptors (SMRT) (Chen and Evans, 1995) and the nuclear receptor corepressor (NCoR) (Horlein et al., 1995). Recent work suggests that SMRT and NCoR play important roles in PPAR␥ action. Data from our group and others have shown that SMRT and
∗
DOI of original article:10.1016/j.mce.2006.08.004. Corresponding author. Tel.: +1 773 834 1012; fax: +1 773 834 0486. E-mail address: [email protected] (R.N. Cohen).
0303-7207/$ – see front matter. Published by Elsevier Ireland Ltd. doi:10.1016/j.mce.2006.10.015
NCoR repress PPAR␥-mediated gene transcription on a subset of target genes (Guan et al., 2005; Yu et al., 2005). By repressing PPAR␥ activity, SMRT and NCoR modulate adipogenesis (Yu et al., 2005). SMRT and NCoR share a similar overall structure (Privalsky, 2004). Each is composed of amino-terminal repressing domains and carboxy-terminal interacting domains (IDs) (Fig. 1A). These IDs (cross-hatched areas in Fig. 1A) mediate interactions with nuclear receptors. SMRT contains two interacting domains, a distal (S1) and a proximal (S2) domain. In contrast, NCoR contains three IDs. Each ID in SMRT and NCoR contains a unique sequence, called a CoRNR box motif (each represented by a gray box in Fig. 1A), that is critical for nuclear receptor–CoR interactions. CoRNR box motifs contains the consensus sequence L–x–x–I/H–I–x–x–x–I/L (also reported as I/L–x–x–I/V–I) (Hu and Lazar, 1999; Nagy et al., 1999; Perissi et al., 1999). The fact that each corepressor ID contains a unique CoRNR box sequence (see below and Fig. 1C) suggests there may be specificity in terms of corepressor recruitment by nuclear receptors. Early work suggested that TR preferred to recruit NCoR whereas RAR␣ preferred SMRT (Cohen et al., 2000). Recent work by Goodson et al. (2005) has suggested that corepressor preference is more complex than previously thought because SMRT consists of two isoforms that differ in their interacting domains. These two isoforms have been termed SMRT␣ and SMRT;
M.M. Sutanto et al. / Molecular and Cellular Endocrinology 267 (2007) 138–143
139
interestingly, SMRT binds TR weakly, but SMRT␣ binds TR more strongly (Goodson et al., 2005). Our prior data suggests that PPAR␥ recruits SMRT specifically through its distal S1 domain, suggesting there is clear ID preference for PPAR␥ (Yu et al., 2005). In the current study, we have studied the mechanisms for PPAR␥–Sl interactions. We mutated residues in SMRT ID1 to resemble those of SMRT ID2. Interestingly, some but not all of these alterations block SMRT recruitment by PPAR␥. Thus, PPAR␥ depends on certain crucial S1 CoRNR box sequences to recruit SMRT. In contrast, mutation of these residues in SMRT ID1 does not block RAR␣–SMRT interactions. These data confirm that SMRT interacts differently with PPAR␥ as opposed to RAR␣, and suggest there may be mechanisms to modulate corepressor recruitment by distinct nuclear receptors. Finally, and in contrast to TR, we show that PPAR␥ interacts preferentially with SMRT over SMRT␣, suggesting another potential mechanism for influencing corepressor recruitment by nuclear receptors. 2. Materials and methods 2.1. Plasmids Upstream activating sequence (UAS)-Luc contains five copies of a Gal4binding site upstream of luciferase. Gal4-SMRT and Gal4-NCoR contain the IDs of SMRT and NCoR, respectively, downstream of Gal4 and have been previously described (Cohen et al., 2000). The Gal4-SMRT mutant constructs were made in Gal4-SMRT. The mutant Gal4-SMRT constructs were created by site-directed mutagenesis using the Quick-Change Mutagenesis Kit (Stratagene). Gal4-SMRT␣ was made from the corresponding region of SMRT␣, which was a generous gift of Martin Privalsky (University of California). The Gal4PPAR␥, Gal4-RAR␣, and Gal4-RXR␣ constructs contain the ligand-binding domains of the nuclear receptors downstream of Gal4. The constructs for VP16PPAR␥2 (Yu et al., 2005) and VP16-RAR␣ (Cohen et al., 2000) contains the full-length nuclear receptors downstream of the VP16 activating domain in the vectors pVP16 and AASV-VP16, respectively. PPRE-Luc has been described previously (Yu et al., 2005).
2.2. Cell culture and transfection 3T3-L1 and CV-1 cells were grown in DMEM supplemented with antibiotics, glutamine, and 10% calf serum or fetal calf serum, respectively. Cells were transfected in 6-well plates using lipofectamine (Invitrogen) in serum-free media, as per manufacturers’ instructions. Three to four hours after transfection, media was changed to fresh serum-free media, and cells extracts were processed the following day for luciferase activity. All transfections were performed in triplicate and repeated at least three times. Fig. 1. Mutation of SMRT ID1 blocks recruitment by PPAR␥: (A) schematic diagram of SMRT and NCoR. SMRT contains two C-terminal interacting domains (ID1 and ID2); NCoR contains three such domains (ID1–ID3). IDs are represented by cross-hatched boxes. Each ID contains a unique CoRNR box sequence (gray boxes see (C) for sequence details); (B) 3T3-L1 cells were transfected with 300 ng UAS-Luc reporter along with 300 ng Gal4-PPAR␥ or Gal4-RXR␣, and 300 ng pSG5-SMRT (or empty pSG5). Cell extracts were obtained 24 h after transfection and assayed for luciferase activity. Data are expressed as relative luciferase activity, where 1.0 represents luciferase activity in the absence of cotransfected pSG5-SMRT; (C) amino acid residues of SMRT and NCoR CoRNR box motifs, based on the consensus sequence of Picard et al.; (D) 3T3L1 cells were transfected with 300 ng UAS-luciferase reporter plasmid along with 300 ng Gal4-SMRT or Gal4-SMRT mutant construct, and 300 ng VP16PPAR␥2 or empty VP16. Data are presented as fold luciferase activity, i.e., the induction of luciferase activity in the presence of the Gal4 construct with VP16-PPAR␥2 compared to that with the empty VP16 vector.
2.3. GST interaction assays GST-S1 and GST-S2 have been previously described (Cohen et al., 2000). GST fusion proteins were expressed in DH5␣ E. coli by induction with 0.1 mM isopropylthio--d-galactosidase at 30 ◦ C. Proteins were isolated after disruption with French Press and purified on Sepharose beads with extensive washing. PPAR␥2 was in vitro translated in reticulocyte lysate (Promega) using T7 polymerase. A 5 ml of 35 S-labeled PPAR␥2 was incubated with GST-corepressor constructs. After extensive washing, bound proteins were eluted by boiling in loading buffer and analyzed by SDS-PAGE.
2.4. RT-PCR mRNA from 3T3-L1 cells was obtained using Micro Fast-Track mRNA Isolation Kit (Invitrogen). RT-PCR for the SMRT isoforms was performed using the
140
M.M. Sutanto et al. / Molecular and Cellular Endocrinology 267 (2007) 138–143
Access RT-PCR system (Promega). The primers used to show relative expression of SMRT␣ and SMRT were the same as previously published by Goodson et al. (2005).
3. Results 3.1. SMRT recruitment to PPARγ-RXR heterodimers is dependent on distinct residues within the SMRT ID1 domain We have previously shown that both SMRT and NCoR downregulate PPAR␥-mediated gene transcription. PPAR␥ binds DNA as a PPAR␥-RXR heterodimer. To determine if SMRT effects are primarily due to the PPAR␥ or RXR components of the heterodimer, we used the Gal4-system. In this way, Gal4PPAR␥ or Gal4-RXR␣ constructs could be tested individually in 3T3-L1 cells. 3T3-L1 fibroblast cells were chosen in these experiments because they have been shown to differentiate into adipocytes under hormonal stimulation. As shown in Fig. 1B, overexpression of SMRT represses Gal4-PPAR␥ transcriptional activity, but is less effective with Gal4-RXR. These data suggest that corepressors do not mediate their effects on PPAR␥ primarily via interactions with the RXR heterodimeric partner. Each corepressor ID contains a unique CoRNR box motif, which is characterized by the sequence L–x–x–I/H–I–x–x–x–I/L (Fig. 1C) (note that in certain papers the CoRNR box motif is represented by the sequence I/L–x–x–I/V–I; since the former version includes the extended helical motif, it will be used here) (Hu and Lazar, 1999; Nagy et al., 1999; Perissi et al., 1999). This is particularly true for the domains SMRT ID1 (S1), SMRT ID2 (S2), NCoR ID1 (N1), and NCoR ID2 (N2). The most proximal NCoR ID, ID3 (N3), contains a variant on this sequence and has no homologous region in SMRT; thus, this manuscript will focus on the other IDs. We have previously shown that PPAR␥ recruits SMRT in 3T3-L1 cells primarily via its S1 domain (Yu et al., 2005). Similarly, the homologous domain of NCoR (N1) is required for full interactions of NCoR and PPAR␥ (data not shown). We have focused on SMRT in the current manuscript since the S1 domain can be altered to resemble S2, and there are only two such domains. Importantly, many CoRNR box amino acid residues are similar in S1 and N1. For example, the residue at the +2 location of the CoRNR box sequence is either a glutamic acid (E) in S1 and N1, or an alanine (A) in S2 and N2. The amino acids at positions +4 and +9 are also conserved. The alanine (A) at position +8 of the S1 and N1 CoRNR box is relatively conserved, in that the homologous residue is a valine (V) in S2 and an isoleucine (I) in N2. These striking similarities between the distal CoRNR box sequences suggest that there are specific amino acid residues in S1 and N1 that are vital for optimal PPAR␥ interactions. Therefore, we mutated amino acids in SMRT ID1 to the corresponding residues in SMRT ID2 to determine which residues are important for full interactions with PPAR␥. To define these interactions between PPAR␥ and mutant SMRT constructs, we performed mammalian two-hybrid experiments in 3T3-L1 cells. 3T3-L1 cells were chosen because they are a well-recognized model for adipogenesis, and are capable of differentiating into adipocytes under the correct hormonal stim-
Fig. 2. Mutation of SMRT blocks SMRT repression of PPAR␥ transcriptional activity: (A) 35 S labeled PPAR␥2 was incubated with SMRT ID constructs, GST-ID1, GST-ID1mut, and GST-ID2. After extensive washing, bound proteins were evaluated by SDS-PAGE upper panel. Mutation of the +4 isoleucine in the ID1 CoRNR box significantly decreases interactions with PPAR␥. The lower panel shows that GST constructs were made in similar amounts, though slightly more GST-ID2 was made than GST-ID1 or GST-ID1mut; (B) each well of 3T3L1 cells was cotransfected with 300 ng of a PPRE-luciferase construct, 300 mg pSG5-PPAR␥2, and 300 ng pSG5, pSG5-SMRT, or pSG5-SMRTmut. Data are expressed as relative luciferase activity, where 1.0 is defined as luciferase activity in the presence of PPRE-Luc, pSG5-PPAR␥2 and empty pSG5 without SMRT or SMRTmut.
ulation. As shown in Fig. 1D, changing the isoleucine at position +4 of SMRT ID1 to histidine completely blocked PPAR␥–SMRT interactions in 3T3-L1 cells. In addition, mutation of glutamic acid to alanine also dramatically decreased interactions, though not to the degree of the isoleucine mutation. In contrast, mutation of alanine to valine did not decrease interactions at all, suggesting that the specific amino acid in this position is not as vital to interactions with PPAR␥, or that either alanine or valine is sufficient for optimal interactions. Next, the isoleucine-to-histidine mutation was inserted into GST-ID1 to determine if this mutation decreases direct recruitment of the isolated interacting domains to PPAR␥. The GST-ID1 and GST-ID2 have been previously described (Cohen et al., 2000). These constructs were incubated with in vitro translated 35 S-labeled PPAR␥2, extensively washed, and analyzed by SDS-PAGE. As shown in Fig. 2A (upper panel), GST-ID1 interacts strongly with PPAR␥, but GST-ID2 interacts weakly if at all with PPAR␥. Interestingly, mutation of the isoleucine at position +4 of SMRT ID1 to the corresponding histidine of ID2 dramatically decreases interactions of ID1 with PPAR␥. The lower panel of Fig. 2A shows that the GST-ID1 and GST-ID1mut constructs were made in similar amounts; although slightly more GST-ID2 was made than GST-ID1, GST-ID2 still did not interact with PPAR␥2. These data confirm that the isoleucine at +4 is vital for SMRT recruitment by PPAR␥ and suggest that the structures of ID1 and ID2 differ in important ways to determine optimal recruitment by nuclear receptors. To determine if the mutation would block SMRT-mediated repression of full-length PPAR␥ transcriptional activity, the
M.M. Sutanto et al. / Molecular and Cellular Endocrinology 267 (2007) 138–143
isoleucine-to-histidine mutation was placed into pSG5-SMRT. These constructs were cotransfected in 3T3-L1 cells with pSG5PPAR␥2 and a PPRE-luciferase reporter. As shown in Fig. 2B, overexpression of pSG5-SMRT decreased PPAR␥-mediated luciferase activity. In contrast, this action was blocked after mutation of the isoleucine of the SMRT ID1 CoRNR box to the corresponding histidine of ID2. These data suggest that func-
141
tional interactions of SMRT with PPAR␥2 are dependent on the ID1 CoRNR box sequence. 3.2. Mutation of the SMRT ID1 isoleucine at position +4 blocks SMRT repression of PPARγ transcriptional activity, but does not block SMRT effects on RARα To determine if the same SMRT ID1 mutation would block interactions with RAR␣, a mammalian two-hybrid assay with VP16-RAR␣ was performed in 3T3-L1 cells. As shown in Fig. 3A, Gal4-SMRT but not Gal4-SMRT I–>H again interacted with VP16-PPAR␥2. In contrast, both Gal4-SMRT and Gal4-SMRT I–>H interacted well with VP16-RAR␣. To compare the functional effects of the SMRT mutation on PPAR␥ and RAR␣ activity, pSG5-SMRT (or pSG5 alone) was cotransfected with Gal4-empty vector or Gal4-PPAR␥ in 3T3-L1 cells. As shown in Fig. 3B, overexpression of SMRT decreased Gal4-PPAR␥ activity. In contrast, overexpression of the SMRT I–>H mutation did not lead to repression. As a control, both constructs did not repress an empty Gal4 construct. We next wanted to determine if the SMRT constructs would lead to repression of RAR. Interestingly, Gal4-RAR␣ exhibited very low levels of transcriptional activity in 3T3-L1 cells, and overexpression of SMRT or the SMRT mutant did not lead an enhancement of repression in these cells. Therefore, we performed the RAR experiments in CV-1 cells, since SMRT has
Fig. 3. A SMRT mutation is specific for interactions with PPAR␥: (A) 3T3L1 cells were transfected with 300 ng UAS-luciferase; 300 ng Gal4-SMRT or Gal4-SMRT (I–>H); 300 ng VP16-PPAR␥2 or VP16-RAR␣ (or the appropriate empty VP16 vector). Data are expressed as fold luciferase activity, i.e., the induction of luciferase activity with VP16-PPAR␥2 (or VP16-RAR␣) compared to that of the empty VP16 vector; (B) 3T3-L1 cells were transfected with 300 ng UAS-luciferase; 300 ng Gal4-PPAR␥ (or empty Gal4); 300 ng empty pSG5, pSG5-SMRT, or pSG5-SMRT (I–>H). Data are expressed as relative luciferase activity, where 1.0 is defined as luciferase activity in the presence of empty pSG5 and empty Gal4 (upper panel) or empty pSG5 and Gal4-PPAR␥ (lower panel); (C) CV-1 cells were transfected as in B, except Gal4-RAR␣ was used at a dose of 100 ng due to differences in transfections between CV-1 and 3T3-L1 cells.
Fig. 4. PPAR␥ prefers to interact with the SMRT isoform: (A) amino acid sequence of the portion of S1 containing differences between SMRT and SMRT␣. Both IDs contain the appropriate S1 CoRNR box as shown. After the CoRNR box, SMRT␣ contains 47 extra amino acids (shown in bold) as compared with SMRT; (B) RT-PCR or 3T3-L1 cells showing the presence of both SMRT and SMRT␣; (C) 3T3-L1 cells were transfected with 300 ng UAS-luciferase; 100 ng Gal4-SMRTt or Gal4-SMRT␣; VP16-PPAR␥2 or VP16-RAR␣ (or the corresponding empty VP16 vector). Data are expressed as fold luciferase activity in the presence of the VP16-PPAR␥2 (pr VP16-RAR␣) construct over that of the empty VP16 vector.
142
M.M. Sutanto et al. / Molecular and Cellular Endocrinology 267 (2007) 138–143
previously been shown to enhance RAR␣-mediated repression in this cell line by other groups (Hauksdottir et al., 2003). In CV1 cells, cotransfection of both pSG5-SMRT and pSG5-SMRT I–>H led to enhanced repression of RAR␣ (Fig. 3C). Thus, SMRT interacts with RAR␣ differently than it does with PPAR␥. 3.3. PPARγ preferentially interacts with the SMRT isoform Recently, Goodson et al. (2005) has shown that SMRT consists of two isoforms, which differ in their sequences just distal to the SMRT ID CoRNR box (Fig. 4A). Our previous work above used SMRT constructs (Cohen et al., 2001; Yu et al., 2005). Thus, we were interested in determining if PPAR␥ could recruit SMRT␣, and how this interaction compared to that with SMRT. Using RT-PCR, we first determined which SMRT isoforms are present in 3T3-L1 cells. In fact, both SMRT␣ and SMRT are present in 3T3-L1 cells, though SMRT appears to be more abundant (Fig. 4B). Next, we performed a mammalian two-hybrid assay to determine whether PPAR␥ could recruit SMRT␣ to a similar degree as SMRT. The portion of SMRT␣ (a generous gift of Martin Privalsky) that was homologous to the SMRT region used previously in Gal4-SMRT was placed downstream of Gal4 to make Gal4-SMRT␣. Gal4-SMRT␣ and Gal4-SMRT were cotransfected with VP16-PPAR␥2 (or VP16 alone). As shown in Fig. 4C, PPAR␥ interacted well with both SMRT isoforms; however, its interactions with SMRT were stronger. In contrast, RAR␣ interacted equally well with both SMRT isoforms in 3T3-L1 cells, similar to previous data in other cell types (Goodson et al., 2005). These data suggest that the PPAR␥ specificity for SMRT ID1 extends to naturally occurring SMRT isoforms that differ in their ID1 sequences.
S1 and S2 are important for PPAR␥ interactions. We found that some, but not all, such residues that are conserved in S1 and N1, but are different in S2, are vital for interactions with PPAR␥. In particular, the isoleucine at position +4 of the S1 CoRNR box sequence was crucial for strong interactions of SMRT with PPAR␥. Mutation of the isoleucine to the corresponding histidine found in S2 blocked SMRT recruitment by PPAR␥, even though all other amino acids in S1 were maintained. When this mutation was introduced into SMRT, SMRT-mediated repression of PPAR␥ was also abolished. The SMRT corepressor is also a major determinant of RAR␣ action. RAR␣ is a nuclear receptor which (similar to PPAR␥) heterodimerizes with RXR. In addition, the SMRT corepressor is also a major determinant of RAR␣ transcriptional activity. RAR␣ binds vitamin A metabolites as ligand (Mangelsdorf et al., 1995). RAR␣ represses gene transcription in the absence of ligand, and this repression is mediated by SMRT and NCoR (Cohen et al., 2000; Hauksdottir et al., 2003). Previous work has shown that RAR␣ interacts better with SMRT than with NCoR, and some of this preference may be mediated by the S2 domain (Cohen et al., 2001). Interestingly, mutation of S1 CoRNR box isoleucine at position +4 to histidine did not decrease RAR␣–SMRT interactions, and the mutant SMRT was still able to repress RAR␣ transcriptional activity. Thus, RAR␣ and PPAR␥ interact differently with SMRT, even though both interactions occur via SMRT IDs. The crystal structure of SMRT complexed to PPAR␥ has not yet been identified. However, the structure of the related receptor PPAR␣ with a portion of SMRT ID1 in the presence of an antagonist has been determined (Xu et al., 2002). Xu et al. showed that the SMRT ID1 CoRNR box sequence forms a three-turn alpha helix that fits into a groove formed by PPAR␣ helices 3, 3 , 4, and 5. This allows the SMRT ID1 residues L1, I5, and
4. Discussion PPAR␥ has been shown to regulate adipogenesis and modulate insulin sensitivity. The corepressor SMRT binds PPAR␥ and represses its transcriptional activity (Cohen, 2006; Guan et al., 2005; Yu et al., 2005). The binding of SMRT to PPAR␥ is regulated at a number of levels, for example, the degree of ligand binding (Yu et al., 2005) and the specific nature of the underlying DNA response element (Guan et al., 2005). The recruitment of SMRT has been shown to affect the ability of TZDs to stimulate adipogenesis (Yu et al., 2005), and it is likely that the interactions between SMRT and PPAR␥ will be important in understanding other PPAR␥-mediated effects such as the regulation of insulin sensitivity. Thus, it is important to understand how SMRT is recruited to PPAR␥. We have focused on the specific amino acid residues in SMRT ID1 (S1). We have previously shown that SMRT binding to PPAR␥ specifically depends on this domain. Furthermore, the CoRNR box sequence of this domain has a number of interesting similarities to that of NCoR ID1, which primarily mediates interactions between NCoR and PPAR␥. Prior data suggests that the other SMRT ID, S2, does not play an important role in SMRT–PPAR␥ interactions (Yu et al., 2005). Thus, we examined the amino acid residues of the S1 CoRNR box sequence to see which residues that differ between
Fig. 5. Schematic diagram of SMRT ID1 CoRNR box complexed to the PPAR␣ ligand binding domain. The structure is derived from PubMed structure PDB 1KKQ (mmdb 18993). The amino acid residues of SMRT ID1 that are included in the structure are represented by spheres (with the CoNR box residues in yellow, and other residues in purple). The core residues (L1, I5, and L9), as well as the mutated residue I4 are indicated with arrows. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)
M.M. Sutanto et al. / Molecular and Cellular Endocrinology 267 (2007) 138–143
L9 to be aligned. In contrast, the residues I4, A8, E2, and R6 are oriented at 90◦ angles to this interface (Fig. 5); however, I4 makes additional hydrophobic contacts along the binding groove (Xu et al., 2002), and is thus in position to allow for differential interactions with nuclear receptors. Goodson et al. (2005) have recently described two SMRT isoforms that differ in S1 sequences just distal to the S1 CoRNR box. We hypothesized that the recruitment of SMRT by PPAR␥ would be affected by differences between these two isoforms. In fact, PPAR␥ recruited SMRT more strongly than SMRT␣ in a two-hybrid experiment. SMRT is the isoform that lacks 47 amino acids present in SMRT␣. This is distinct from RAR␣, which recruited both SMRT isoforms equally well. This contrasts with the TR, which has been shown to bind SMRT␣ better than SMRT (Goodson et al., 2005). Our data suggests that 3T3L1 cells contain both SMRT isoforms. It is possible that one or the other SMRT isoform might be preferentially expressed under certain circumstances, which might allow differing transcriptional effects in the cell. Thus, it might be possible to alter adipogenesis via changing the proportional amounts of SMRT versus SMRT␣ in the cell. Although our data are limited to PPAR␥–SMRT isoform interaction assays, we anticipate that these differences in interactions will be reflected in the functional ability of the SMRT isoforms to repress PPAR␥ transcription. We have not yet been able to perform those experiments, since our full-length SMRT constructs are derived from different species, and are thus beyond the scope of the current manuscript. PPAR␥ has been shown to have play multiple roles in vivo. Its best characterized function is as the central regulators of adipogenesis (Lehrke and Lazar, 2005). More recently, PPAR␥ has also been implicated in other tissues, such as the immune system (Pascual et al., 2005), muscle (Hevener et al., 2003; Norris et al., 2003), liver (Matsusue et al., 2003), and the gastrointestinal tract (Auwerx, 2002). TZDs serve as ligands for PPAR␥. Selective PPAR␥ modulators (SPPARMs) may be developed with differing profiles of transcriptional activation (Auwerx, 2002). SPPARMs enhance coactivator recruitment in certain tissues or on certain promoters, and corepressor recruitment on others. Thus, it is important to understand those factors that lead to corepressor recruitment by PPAR␥. A further understanding of how PPAR␥ recruits SMRT, and how PPAR␥ differs from other nuclear receptors in this fashion, will eventually allow us to define ways to modulate SMRT effects on adipogenesis and insulin sensitivity. Acknowledgements This work was supported by R21 DK072397. We gratefully acknowledge Martin Privalsky (SMRT␣) and J. Don Chen and Ronald Evans (SMRT) for plasmids. We thank Katie Markan for technical assistance. References Auwerx, J., 2002. Nuclear receptors. I. PPAR gamma in the gastrointestinal tract: gain or pain? Am. J. Physiol. Gastrointest. Liver Physiol. 282, G581–G585.
143
Chen, J.D., Evans, R.M., 1995. A transcriptional co-repressor that interacts with nuclear hormone receptors. Nature 377, 454–457. Cohen, R.N., 2006. Nuclear receptor corepressors and PPARgamma. Nucl. Recept. Signal 4, e003. Cohen, R.N., Brzostek, S., Kim, B., Chorev, M., Wondisford, F.E., Hollenberg, A.N., 2001. The specificity of interactions between nuclear hormone receptors and corepressors is mediated by distinct amino acid sequences within the interacting domains. Mol. Endocrinol. 15, 1049–1061. Cohen, R.N., Putney, A., Wondisford, F.E., Hollenberg, A.N., 2000. The nuclear corepressors recognize distinct nuclear receptor complexes. Mol. Endocrinol. 14, 900–914. Goodson, M.L., Jonas, B.A., Privalsky, M.L., 2005. Alternative mRNA splicing of SMRT creates functional diversity by generating corepressor isoforms with different affinities for different nuclear receptors. J. Biol. Chem. 280, 7493–7503. Guan, H.P., Ishizuka, T., Chui, P.C., Lehrke, M., Lazar, M.A., 2005. Corepressors selectively control the transcriptional activity of PPARgamma in adipocytes. Gene Dev. 19, 453–461. Hauksdottir, H., Farboud, B., Privalsky, M.L., 2003. Retinoic acid receptors beta and gamma do not repress, but instead activate target gene transcription in both the absence and presence of hormone ligand. Mol. Endocrinol. 17, 373–385. Hevener, A.L., He, W., Barak, Y., Le, J., Bandyopadhyay, G., Olson, P., Wilkes, J., Evans, R.M., Olefsky, J., 2003. Muscle-specific Pparg deletion causes insulin resistance. Nat. Med. 9, 1491–1497. Horlein, A.J., Naar, A.M., Heinzel, T., Torchia, J., Gloss, B., Kurokawa, R., Ryan, A., Kamei, Y., Soderstrom, M., Glass, C.K., Rosenfeld, M.G., 1995. Ligand-independent repression by the thyroid hormone receptor mediated by a nuclear receptor co-repressor. Nature 377, 397–404. Hu, X., Lazar, M.A., 1999. The CoRNR motif controls the recruitment of corepressors by nuclear hormone receptors. Nature 402, 93–96. Lehrke, M., Lazar, M.A., 2005. The many faces of PPARgamma. Cell 123, 993–999. Mangelsdorf, D.J., Thummel, C., Beato, M., Herrlich, P., Schutz, G., Umesono, K., Blumberg, B., Kastner, P., Mark, M., Chambon, P., Evans, R.M., 1995. The nuclear receptor superfamily: the second decade. Cell 83, 835– 839. Matsusue, K., Haluzik, M., Lambert, G., Yim, S.H., Gavrilova, O., Ward, J.M., Brewer Jr., B., Reitman, M.L., Gonzalez, F.J., 2003. Liver-specific disruption of PPARgamma in leptin-deficient mice improves fatty liver but aggravates diabetic phenotypes. J. Clin. Invest. 111, 737–747. Nagy, L., Kao, H.Y., Love, J.D., Li, C., Banayo, E., Gooch, J.T., Krishna, V., Chatterjee, K., Evans, R.M., Schwabe, J.W., 1999. Mechanism of corepressor binding and release from nuclear hormone receptors. Gene Dev. 13, 3209–3216. Norris, A.W., Chen, L., Fisher, S.J., Szanto, I., Ristow, M., Jozsi, A.C., Hirshman, M.F., Rosen, E.D., Goodyear, L.J., Gonzalez, F.J., et al., 2003. Musclespecific PPARgamma-deficient mice develop increased adiposity and insulin resistance but respond to thiazolidinediones. J. Clin. Invest. 112, 608–618. Pascual, G., Fong, A.L., Ogawa, S., Gamliel, A., Li, A.C., Perissi, V., Rose, D.W., Willson, T.M., Rosenfeld, M.G., Glass, C.K., 2005. A SUMOylationdependent pathway mediates transrepression of inflammatory response genes by PPAR-gamma. Nature 437, 759–763. Perissi, V., Staszewski, L.M., McInerney, E.M., Kurokawa, R., Krones, A., Rose, D.W., Lambert, M.H., Milburn, M.V., Glass, C.K., Rosenfeld, M.G., 1999. Molecular determinants of nuclear receptor–corepressor interaction. Gene Dev. 13, 3198–3208. Privalsky, M.L., 2004. The role of corepressors in transcriptional regulation by nuclear hormone receptors. Annu. Rev. Physiol. 66, 315–360. Xu, H.E., Stanley, T.B., Montana, V.G., Lambert, M.H., Shearer, B.G., Cobb, J.E., McKee, D.D., Galardi, C.M., Plunket, K.D., Nolte, R.T., et al., 2002. Structural basis for antagonist-mediated recruitment of nuclear co-repressors by PPARalpha. Nature 415, 813–817. Yu, C., Markan, K., Temple, K.A., Deplewski, D., Brady, M.J., Cohen, R.N., 2005. The nuclear receptor corepressors NCoR and SMRT decrease peroxisome proliferator-activated receptor gamma transcriptional activity and repress 3T3-L1 adipogenesis. J. Biol. Chem. 280, 13600– 13605.