Selective Coactivator Interactions in Gene Activation by SREBP-1a and -1c

Abstract

Requisite levels of intracellular cholesterol and fatty acids are maintained in part by the sterol regulatory element binding proteins (SREBPs). Three major SREBP isoforms exist; SREBP-1a and SREBP-1c are expressed from overlapping mRNAs, whereas SREBP-2 is encoded by a separate gene. The active forms of SREBP-1a and SREBP-1c differ only at their extreme N termini; SREBP-1c lacks 28 aa present in SREBP-1a and instead contains 4 unique aa of its own. While the SREBP-1a and -1c isoforms differentially activate transcription, the molecular basis of this difference is unknown. Here we define the differences between these proteins that confer the enhanced activity of SREBP-1a and demonstrate that this enhancement is a direct result of its avid binding to the coactivator CREB binding protein (CBP) and the mammalian mediator complex. While previous work determined that the C/H1 zinc finger and KIX domains of CBP bind to SREBP-1a, we provide evidence that the interaction with C/H1 is important for gene activation. We further show that the association between the activation domain of SREBP-1 and mediator is through aa 500 to 824 of DRIP150. Finally, we demonstrate the recruitment of mediator to an SREBP-responsive promoter in a sterol-dependent manner.

Cholesterol homeostasis in mammalian cells is maintained in part by the basic-helix-loop-helix (bHLH) family of transcription factors called the sterol regulatory element binding proteins (SREBPs) (3). These proteins are made as precursors that contain two membrane-spanning domains that retain them in the membrane of the endoplasmic reticulum and nuclear envelope. When cells are depleted of sterols, the precursor proteins are cleaved twice by distinct proteases, releasing the mature transcriptionally active portion from its membrane tether (41). The processed SREBPs then translocate to the nucleus as preformed dimers (25), where they activate a number of target genes involved in cholesterol and fatty acid metabolism through binding to sterol regulatory elements in the promoters of target genes (2, 10, 13, 17, 21, 26, 55).

There are three major SREBP isoforms; SREBP-1a and SREBP-1c are products of the same gene, whereas SREBP-2 is encoded by a second distinct genetic locus (20, 29). mRNAs for the two SREBP-1 isoforms are transcribed from different promoters, and each one contains an exclusive 5′-terminal exon that is spliced to a shared second exon and the proteins are identical from this point forward (20, 55). The resulting SREBP-1c amino terminus contains 4 unique amino acids (aa) but lacks 28 others that are only present in SREBP-1a (Fig. 1A and C). The unique region in SREBP-1a is part of a potent activation domain. The related protein SREBP-2 is very similar to SREBP-1a in this region (Fig. 1C) and is also a potent activator of gene expression (20, 29).

FIG. 1.

Select SREBP isoforms that interact with CBP and the residues required for this interaction. (A) Schematic representation of the mature SREBP-1 isoforms, SREBP-1a and SREBP-1c. The two isoforms are alternative splice products of the gene for SREBP-1 and differ in the first exons, which encode the N termini of their activation domains. SREBP-1a has a much longer exon 1 than SREBP-1c and contains residues that allow it to interact with the mediator (Med) complex and CBP. SREBP-1c is a much weaker activator than SREBP-1a and interacts weakly with mediator and CBP. (B) Lysates from E. coli expressing GST (G) or GST fusions containing the indicated SREBPs were bound to glutathione-agarose beads, washed, and incubated in the absence (−) or presence (+) of HeLa cell nuclear extracts (N.E.). Bound proteins were resolved by SDS-PAGE and visualized by immunoblot analysis with antisera directed against CBP and the A subunit of NF-Y. Input HeLa cell nuclear extract was analyzed in lane 1 (In). (C) Alignment of the N-terminal amino acids from human SREBP-2, -1c, and -1a. The residues with the thick overline correspond to the predicted α-helical region evaluated by Cardinaux et al. (5). The underlined regions labeled Δ1 and Δ2 show residues retained in these deletion mutants. Amino acids in either SREBP-1a or -1c that were mutated are both in bold and underlined. (D) The indicated regions of SREBP-1a were expressed as GST fusions and purified from E. coli. Equivalent amounts (18 μM final concentration) were bound to glutathione-agarose and incubated with 1 mg of nuclear extract from HeLa cells. The samples were washed, and bound material was eluted and resolved by SDS-PAGE. Interacting proteins were visualized by immunoblot analysis with an antibody directed against CBP. HeLa cell nuclear extract (In; 12 μg) was directly analyzed in lane 1. Lanes 2 to 10 show equivalent amounts of eluted material from columns bound with GST (G) or the indicated GST-SREBPs.

Another defining feature that is common to all three SREBPs is that they must work with ubiquitous DNA binding coregulatory factors that recognize neighboring sites in target promoters. These coregulatory proteins exhibit some promoter specificity. For example, the single coregulator for the low-density lipoprotein (LDL) receptor promoter is Sp1 (42) whereas at least two coregulators, CREB and nuclear factor Y (NF-Y), are required for 3-hydroxy-3-methylglutaryl (HMG) coenzyme A (CoA) synthase transcriptional activation (9).

Because whole-genome sequencing efforts have uncovered the surprising observation that there are only approximately 30,000 individual gene loci in mammals, differential splicing and alternative promoter usage are likely to contribute significantly to overall genome complexity (53). In this light, it has become increasingly important to define the individual and common functions mediated by proteins that are encoded from overlapping mRNAs from a single gene. The precise roles of the related SREBP isoforms have not yet been firmly established. A series of reports have explored their differential expression and activity, as well as the regulatory consequences of their aberrant expression (41, 44, 46-50). Taken together, these studies indicate that SREBP-1 preferentially activates fatty acid metabolism genes while SREBP-2 preferentially activates cholesterol biosynthesis genes. Further, SREBP-1a is a potent activator of gene expression whereas, in contrast, SREBP-1c is relatively weak.

The amino-terminal 50 aa of both SREBP-1a and SREBP-2 interact with the CREB binding protein (CBP) and p300 family of coactivator proteins (33). This region of both SREBP-1a and -2 is predicted to have an α-helical domain and is similar in sequence to the kinase-inducible domain (KID) of CREB (5). The KID is a substrate for the cAMP-dependent protein kinase and after phosphorylation it binds to the KIX domain of CBP and p300, directly linking cAMP signaling to transcriptional activation (15).

CBP and p300 have defined roles in many other cellular processes, and they have multiple functional domains throughout their coding sequences that interact with several transcriptional activators and transcription factor complexes (6, 7, 11, 22). Both the C/H1 zinc finger and KIX domains of CBP interact with the SREBP-1a activation domain, and there is independent evidence that each one participates in SREBP-dependent gene activation (12, 30). The interaction of a single activation domain with multiple domains of CBP is not a novel phenomenon. For example, Ets family members Ets-1 and -2 interact with both the C/H1 and C/H3 zinc finger domains of CBP (54). However, the relative contribution of the distinct interactions of Ets-1 and -2 with multiple domains of CBP has not been determined. Further, a direct comparative analysis of the distinct contributions of SREBP interacting with the C/H1 and KIX domains of CBP has not been reported.

The SREBP-1a activation domain also interacts with the mammalian mediator complex, which has been characterized independently in several different laboratories and given several different names, including the activator-recruited cofactor complex (31), the vitamin D receptor-interacting protein (DRIP) complex (37), and the thyroid hormone receptor-associated protein coactivator complex (14). Despite some variation in the subunit composition in these different reports, the megadalton-sized complex is homologous to the yeast mediator and plays a critical role in the activation of transcription through interaction with a broad range of sequence-specific DNA binding proteins (35). There is also evidence that different forms of the mediator complex exist that have a slightly altered subunit composition (28). Specific transcription factors interact with distinct subunits of the mediator complex, suggesting that it may stimulate gene expression in different ways, depending on how these factors are recruited to a specific promoter. The subunit(s) that interacts with SREBPs has not been reported.

To further understand the mechanistic basis of how SREBP-1a and -1c differentially activate transcription, we sought to characterize the interactions between SREBPs and non-DNA binding coactivators. Here we report the characterization and comparative analysis of SREBP-1a and -1c, including several mutant versions of each protein. In total, these studies show a good correlation between binding of the C/H1 domain and promoter activation while there is not a strong association between mutations that affect promoter activation and those that affect KIX binding. In addition, SREBPs can activate transcription in mouse embryonic fibroblasts (MEFs) derived from animals containing mutations in the KIX domain of either CBP or p300. We also show that the 150-kDa subunit of the mediator complex binds directly to the SREBP-1a activation domain, specifically through aa 500 to 824, with a specificity similar to that of the CBP C/H1 domain. Cotransfection of a cDNA expressing the 150-kDa subunit, but not several others, stimulates SREBP-dependent activation, and we demonstrate that the mediator is recruited to the endogenous SREBP-responsive HMG CoA reductase promoter under sterol depleted conditions.

In comparison to that of SREBP-1a, the truncated activation domain of SREBP-1c is predicted to lack a portion of the region required to interact with CBP and the mediator complex. We also show that in contrast to SREBP-1a, SREBP-1c interacts very weakly with both CBP and the DRIP150 subunit. The weak binding that is observed requires the SREBP-1c unique amino-terminal domain, which is also important for its ability to stimulate gene expression. These results are sufficient to explain the relatively weak transcriptional activity for SREBP-1c in cultured cells.

MATERIALS AND METHODS

Cells and media.

Human embryonic kidney 293T cells were obtained from C. Walsh (University of California, Irvine) and maintained in Dulbecco's modified Eagle's medium (Irvine Scientific) with 10% fetal bovine serum (Omega Scientific) and additives purchased from Life Technologies, Inc. When endogenous SREBPs were induced, fetal bovine serum was omitted from the above medium and replaced with 0.1% delipidated bovine serum albumin (BSA; A-3803; Sigma) and 1% insulin-transferrin-selenite (I-1884; Sigma). This medium was referred to as defined serum-free medium (DSFM). Where indicated, cholesterol and 25-hydroxycholestrol (both from Sigma) were added to the DSFM from stock solutions such that the final concentrations were 12 and 1 μg/ml, respectively. FLAG-DRIP150 HeLa cells were a kind gift from C. Horvath (Mt. Sinai School of Medicine) and were maintained as previously described (24). Wild-type, CBP^K/K, and p300^K/K MEFs were generous gifts from P. Brindle (St. Jude Children's Research Hospital) and were maintained and used in transient transfections as previously described (23). These cells were subsequently immortalized with simian virus 40 T antigen for use as a source of endogenous CBP.

Plasmids and plasmid construction. (i) Mature SREBP-1 and mutations.

Full-length mature SREBP-1a (human, aa 1 to 490) was cloned into pCDNA3.1+ (Invitrogen) with two copies of the FLAG epitope sequence (Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys-Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys) cloned in frame (2XFLAGpCDNA3.1+) at the amino terminus of the SREBP coding sequence. Deletion of aa 2 to 13 (Δ1) and single point mutations in the Δ1 context were introduced by PCR-based mutagenesis. The positions of the mutations are described in Results and in the figure legends. Mature SREBP-1c and SREBP-1c Δ2-5 were cloned into 2XFLAGpCDNA3.1+ similar to SREBP-1a. SREBP-1a aa 1 to 490 were also cloned into a pCDNA3.1+ with two c-Myc epitopes (Glu-Gln-Lys-Leu-Ile-Ser-Glu-Glu-Asp-Leu-Glu-Gln-Lys-Leu-Ile-Ser-Glu-Glu-Asp-Leu) cloned in frame (2XMycpCDNA3.1+) upstream of the SREBP coding sequence. The mature forms of SREBP-1a, -1c, and -2 were also cloned into pGEX-2T (Pharmacia) as described previously (9).

(ii) SREBP hybrids.

The amino-terminal 100 aa of SREBP-1a or the amino-terminal 75 aa of SREBP-1c were fused to the yeast GAL4 DNA binding domain (DBD; aa 1 to 147) by cloning into pFA-CMV (Stratagene). Mutants of each isoform were constructed by PCR-based mutagenesis with the appropriate DNA oligonucleotides and are described in the figure legends and in Results. The same SREBP-1a or SREBP-1c activation domain derivatives were cloned into pGEX-2T for expression as glutathione S-transferase (GST) fusions. Additional mutants generated in the GST context include 1aΔ1 (aa 2 to 13 deleted), 1aΔ2 (aa 2 to 25 deleted), and 1aΔ50 (aa 51 to 100 deleted). The single and double point mutations in the Δ1 context are illustrated in Fig. 1C. The SREBP coding sequence in all constructs was confirmed independently by DNA sequencing. GAL4VP16 was generated in a similar manner with the 76-aa activation domain of the herpes simplex virus VP16 protein.

(iii) DRIP and CBP.

Cytomegalovirus (CMV)-based expression vectors encoding DRIP subunits 240, 205, 150, and 77 were described earlier (36, 37). These were used in DNA transfection assays or in in vitro transcription-translation reactions as described in Results. Full-length DRIP150 was expressed in pCDNA3, while DRIP150 fragments containing aa 1 to 824 and aa 500 to 824 were subcloned into 2XFLAGpCDNA3.1+ as described above for FLAG-SREBP-1a fusions. Murine CBP fragments aa 1 to 451 (C/H1 domain) and aa 452 to 721 (KIX domain) were similarly cloned into 2XFLAGpCDNA3.1+.

Protein-protein interactions. (i) GST-SREBP activation domain protein expression and purification.

GST-tagged SREBP recombinant proteins were expressed and processed into crude lysates as described earlier (10), except that bacterial expression strain BL21 was used. Crude lysates were bound to glutathione-agarose (Sigma) for 2 h at 4°C with rotation. Nonbound proteins were removed, and the resin was washed extensively with HEGN-0.1 (50 mM HEPES [pH 7.6], 0.1 mM EDTA [pH 8.0], 10% glycerol, 0.1% NP-40, 100 mM KCl) plus protease inhibitors (PI; leupeptin at 1 μg/ml, pepstatin at 0.7 μg/ml, and phenylmethylsulfonyl fluoride at 200 μM) plus dithiothreitol (1 μM). GST proteins were eluted with HEGN-0.1 plus 10 mM glutathione (Sigma) and dialyzed to remove the glutathione. All fusion protein concentrations were estimated by Coomassie blue staining following sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) analysis. Protein band intensities were compared to serial dilutions of a common standard.

(ii) GST fusion protein binding.

For binding with crude HeLa cell nuclear extracts, equivalent amounts (18 μM) of GST-SREBP activation domain fusion proteins were bound to 100 μl of glutathione-agarose (50:50 bead slurry) in a total volume of 300 μl (volumes were equalized with HEGN-0.1 plus PI) for 2 h at 4°C (all reactions were performed with Micro Bio-Spin chromatography columns [Bio-Rad]). Meanwhile, HeLa cell nuclear extract (cell pellets were purchased from the Cell Culture Center, and nuclear extract was prepared as described by Dignam et al. [8]) was precleared over GST-bound glutathione-agarose (1 mg of HeLa cell nuclear extract was used per GST binding reaction, and each preclearing was performed with a total volume of 300 μl) for 30 min at 4°C. Following binding of GST fusions to glutathione-agarose, unbound proteins were removed and the beads were washed three times with HEGN-0.1 (300-μl wash volume). Precleared HeLa cell nuclear extracts were added to the GST-SREBP-bound beads and incubated for 4 h at 4°C. Beads were pelleted and washed six times with HEGN-0.1 plus PI (300-μl wash volume). Bound proteins were eluted by incubation with HEGN-0.1 plus 10 mM glutathione (60 μl) at 4°C for 1 h. Portions of the eluted material were analyzed by SDS-PAGE and immunoblotting with an antibody directed against CBP (sc-369; Santa Cruz) and NFY-A (Rockland). For analysis of interactions with in vitro-translated proteins, equivalent molar amounts of GST-SREBPs (50-μg equivalents) were incubated with ³⁵S-labeled proteins (wild-type DRIP150, DRIP150 aa 500 to 824, the CBP C/H1 domain, and the CBP KIX domain) generated by in vitro transcription and translation with the TNT coupled rabbit reticulocyte system (Promega). Binding reactions were performed with HEGN-0.1 plus BSA at 1 mg/ml and PI for 2 h at 4°C. Glutathione-agarose (25 μl of a 50:50 bead slurry) was then incubated with the in vitro-translated products for an additional 2 h at 4°C. Beads were then pelleted, washed with HEGN-0.1 plus PI (three times, 300-μl wash volume), and boiled in 2× SDS sample buffer plus β-mercaptoethanol. Samples were resolved by SDS-PAGE, and the gel was stained with Coomassie blue to ensure that equivalent amounts of GST proteins were eluted. The gel was then soaked in Amplify (Amersham) and dried, and interacting proteins were visualized by fluorography. For binding reactions performed in the presence of spermine, 1 mM spermine was added during the first incubation and maintained throughout the assay. Interaction of GST-SREBP-1a or -1c with nuclear extracts from wild-type and CBP^K/K MEFs was performed as described above, except that nuclear extracts were made as described in reference 1 and 0.4 M KCl was substituted for 0.4 M LiCl in buffer B. Extracts were then diluted to 0.2 M KCl with HEGN (no KCl) plus PI for binding to GST-SREBP-1-bound beads. Following washing with HEGN-0.1 plus PI, interacting proteins were eluted by boiling in 2× SDS sample buffer plus β-mercaptoethanol. Portions of the eluted material were analyzed by SDS-PAGE and immunoblotting with antibodies directed against CBP (no. 06-294; Upstate) and GST (Sigma).

(iii) SREBP-DRIP150 coimmunoprecipitation.

293T cells were plated on 10-cm-diameter dishes at 8 × 10⁶ cells/plate on day 0. Cells were transfected with Lipofectamine reagent (Invitrogen) on day 1 in accordance with the manufacturer's suggested protocol and refed with complete medium (described above) at 5 h posttransfection. Cells were harvested on day 3 in HEGN-0.1 plus PI (1 ml/plate) and lysed on ice for 10 min, and cell debris were pelleted at 13,500 × g and 4°C for 5 min. Sepharose beads covalently coupled with an antibody to the c-Myc epitope (30 μl of a 50:50 slurry; Covance) were blocked in HEGN-0.1 plus BSA at 1 mg/ml and PI for 1 h at 4°C in Micro Bio-Spin chromatography columns on a rotator. The blocking buffer was removed, 300 μl of each cell lysate (plus BSA at a final concentration of 1 mg/ml) was added to the beads, and the mixture was incubated for an additional 1 h at 4°C. The beads were then washed with HEG + 0.1 plus PI (same buffer as above, no NP-40) and transferred to a tube. Bound material was eluted by adding 2× SDS sample buffer plus β-mercaptoethanol and boiling the mixture. The eluates were resolved by SDS-PAGE, and interacting proteins were visualized by immunoblotting with either anti-c-Myc (sc-789; Santa Cruz) or anti-FLAG (F-3165; Sigma) antibodies.

Transient transfections. (i) Protein expression studies.

293T cells were plated on day 0 in six-well dishes at 10⁶ cells/well and transfected on day 1 with Lipofectamine reagent (described above). Cells were harvested on day 2 in lysis buffer A (50 mM HEPES [pH 7.5], 150 mM NaCl, 0.1 mM EDTA [pH 8], 10% glycerol, 1% Triton X-100) plus PI and lysed on ice for 10 min, and cell debris were pelleted at 13,500 × g and 4°C for 5 min. Extracts were assayed for protein concentration with Bradford reagent (Bio-Rad) and for β-galactosidase activity as described previously (52). Protein concentrations were normalized to β-galactosidase values, and the proteins were resolved by SDS-PAGE and visualized by immunoblotting with antibodies to the Gal4 DBD (sc-510; Santa Cruz) and the FLAG epitope.

(ii) Promoter activation studies.

293T cells were plated in six-well dishes at 3 × 10⁵ cells/well on day 0. On day 1, the cells were transfected by calcium phosphate coprecipitation. Individual plasmids used in each transfection are noted in the figure legends. On day 2 (16 h posttransfection), the cells were washed two times with 1× phosphate-buffered saline and refed with complete medium. Cells were harvested on day 3 and analyzed for luciferase and β-galactosidase activities by the standard methods cited above. Each experimental datum point is based on duplicate measurements and is representative of several independent transfections. In activation assays performed with endogenously encoded SREBPs and transfected individual DRIP subunits, 293T cells were plated and transfected as described above and on day 2 cells were refed with DFSM minus sterols or DFSM plus sterols (plus cholesterol and 25-hydroxycholesterol [final concentrations, 12 and 1 μg/ml, respectively]). Cells were harvested on day 3 as described above. Because the DRIP subunits are likely to be involved in activation of the CMV promoter as well, in these experiments, both luciferase and β-galactosidase values were normalized to the total protein amounts (concentrations were determined by the Bradford assay) and each was independently represented relative to the total protein concentration. CV-1 cells were plated in six-well dishes at 8.5 × 10⁴ cells/well on day 0 and transfected by calcium phosphate coprecipitation as described for 293T cells. MEFs were transfected as described in reference 23 and refed either normal medium or DFSM minus or plus sterols. Cells were harvested at 24 h posttransfection and assayed for luciferase and β-galactosidase activities as described above.

ChIP analysis.

Chromatin immunoprecipitation (ChIP) analysis was performed essentially as previously described (32), with the following minor modifications. Endogenous SREBP expression was induced for 24 h in these cells with the DSFM described above (minus or plus sterols). The FLAG-DRIP150 HeLa cells (24) were treated with formaldehyde for 10 min at room temperature, and sonication was performed in 10-s bursts (14 times per sample) on a Heat Systems Ultrasonics sonicator (model w-220F) at setting 6 with a microtip. Anti-FLAG M2 antibody (125 μg per reaction; Sigma) and beads were washed four times with 500 μl of ice-cold buffer B. For the PCRs, 1 μl of DNA from the FLAG precipitation was used and PCR oligonucleotides that amplify a 250-bp fragment from the human HMG CoA reductase promoter (sequence available on request) were used. All PCRs were performed in triplicate at 32 cycles and monitored for amplification (see Fig. 7) to ensure that the signals were in the linear range of the PCR. To normalize the signal intensities of PCRs performed on FLAG antibody-immunoprecipitated reactions, PCRs on material recovered from immunoprecipitations without FLAG antibody added were also performed (no antibody, Fig. 7B) and the quantitated signal intensities of these reactions were subtracted from the signal intensities of the PCRs of the DNA from the anti-FLAG antibody-immunoprecipitated samples. To detect FLAG-DRIP150 and SREBP-1, endogenous SREBP-1 expression was induced as described earlier and MG132 (50 mM in dimethyl sulfoxide [DMSO]; Calbiochem) was added to select samples at a final concentration of 25 μM for the final 12 h of the 24-h induction period (an equivalent amount of DMSO was added to the samples not treated with MG132). The cells were harvested and processed into nuclear extracts as previously described (1). Nuclear extracts (70 μg) were resolved by SDS-PAGE, and proteins were detected by immunoblotting with antibodies directed against the FLAG epitope (Sigma) and SREBP-1 (immunoglobulin G2A4; sc-13551; Santa Cruz).

FIG. 7.

Sterol-responsive SREBP-mediated recruitment of DRIP150 to the endogenous HMG CoA reductase promoter. HeLa cells stably expressing FLAG-DRIP150 were fed for 24 h with either DFSM minus sterols to induce endogenous SREBP expression or DSFM plus sterols to suppress endogenous SREBP expression. The cells were then formaldehyde cross-linked, the nuclei were isolated and sonicated, and the resulting extracts were precipitated with an antibody directed against the FLAG epitope as described in Materials and Methods. (A) Serial dilutions of input DNA from induced samples were analyzed by PCR (triplicate reactions) with primers designed to the HMG CoA reductase promoter. Signal intensities were plotted as a function of the input DNA to determine the linear range of the PCR response. (B) PCRs of input DNA (3.3% of induced [minus sterols] and suppressed [plus sterols]) and PCRs with samples following immunoprecipitation (IP) with either no antibody or anti-FLAG antibody are shown at the top of the graph as indicated (1% of the material recovered by immunoprecipitation was used for the PCR). The overall fold induction (10.9) from suppressed to induced samples is indicated. (C) Non-cross-linked nuclear extracts (N.E.) from FLAG-DRIP150 HeLa cells treated as described above, except that MG132 (25 μM) (or DMSO for other samples) was included in the final 12 h of SREBP induction. Nuclear extracts (70 μg) were resolved by SDS-PAGE and immunoblotted with antibodies directed against the FLAG epitope and SREBP-1 (immunoglobulin G2A4). The asterisk denotes a band nonspecifically cross-reacting with the indicated antibody.

RESULTS

SREBP-1a and -1c have separate promoters that ultimately result in two proteins with alternative amino termini, as discussed in the Introduction and shown in Fig. 1A. The remaining coding information is the same in the two proteins; therefore, the reasons why SREBP-1a is a potent activator relative to SREBP-1c must be functional differences at their amino termini. This region of SREBP-1a contains its potent activation domain that interacts with the coactivators CBP and p300 and the multisubunit human mediator complex (also known as the activator-recruited cofactor, DRIP, and the thyroid hormone receptor-associated protein).

To compare the interactions between the SREBP-1a and -1c activation domains and coactivators, the coding sequences for the entire mature form of each SREBP-1 isoform and SREBP-2 were expressed in Escherichia coli as GST fusion proteins. The recombinant proteins were bound to glutathione-agarose beads and incubated with nuclear extract from HeLa cells, and interacting proteins were eluted, resolved by SDS-PAGE, and detected by immunoblotting with specific antibodies. CBP was present in the starting nuclear extract (Fig. 1B, lane 1), and it was present in eluates from the SREBP-1a and SREBP-2 fusion proteins, but not in the eluate from the column containing only the GST protein (Fig. 1B, compare lanes 3, 5, and 9). In contrast, the interaction with SREBP-1c was very weak (lane 7). In a previous study (10), the bHLH-leucine zipper region of SREBP-1 was shown to interact with NF-Y, an interaction that should be common to both SREBP-1 proteins. When the same immunoblot was reacted with an antibody to the A subunit of NF-Y, the strength of the interaction with SREBP-1c was indistinguishable from that observed for the other two SREBPs (Fig. 1B, lanes 5, 7, and 9). These results indicate that the unique N terminus of SREBP-1c interacts weakly with CBP, but interactions occurring in the conserved bHLH-leucine zipper region are retained.

The amino termini of the three human SREBP isoforms are aligned in Fig. 1C. The heavy bold overline marks a stretch of amino acids that were noted in a previous study as being similar to the two-helix-containing KID of CREB (5). In an effort to discern the mechanism for the kinase-inducible interaction between CREB and CBP compared to the proposed constitutive interaction between SREBP-1a and CBP, Cardinaux et al. (5) swapped residues in CREB with the corresponding amino acids in SREBP-1a and vice versa. The binding results were consistent with the similarly positioned amino acids being important for interactions with CBP in each protein. To extend these results and specifically analyze amino acid residues conserved between SREBP-1a and SREBP-2, we fused the amino-terminal activation domain of SREBP-1a to GST and made mutations in this context focusing primarily on residues that are absent in SREBP-1c. Several mutations were introduced into SREBP-1a, the most informative of which were the two deletions and four substitution mutations underlined in Fig. 1C.

These GST-SREBP-1a fusion proteins were bound to glutathione-agarose, and interactions with CBP were evaluated as shown in Fig. 1D. CBP was detectable in the starting crude extract (lane 1), and it bound equivalently to constructs 1a (containing aa 1 to 100 of SREBP-1a), Δ50 (aa 51 to 100 deleted), and Δ1 (lanes 3 to 5). However, deletion of an additional 12 aa as in the Δ2 mutant resulted in significantly lower CBP binding. Proteins bearing point mutations D22-A and ED30/31-AA interacted efficiently with CBP (lanes 7 and 10), whereas mutation of either the conserved leucine (L26-A) or the conserved DI28/29-AA generated mutant proteins that failed to interact (lanes 8 and 9).

To determine how these mutations affect promoter activation by SREBPs, we introduced the same mutations into the wild-type mature SREBP-1a coding sequence and compared the abilities of these proteins to activate two different SREBP target genes in a transfection assay (Fig. 2A and B). For this analysis, we chose the promoters for HMG CoA synthase and the LDL receptor because the SREBPs stimulate these two promoters with different auxiliary DNA binding proteins. In HMG CoA synthase, the auxiliary proteins are NF-Y and CREB (9) while Sp1 is the singular coregulatory protein that activates the LDL receptor promoter along with the SREBPs. In both cases, the Δ1 mutant activated as efficiently as wild-type SREBP-1a over the entire concentration range evaluated. The mutants that were still able to bind endogenously expressed CBP in the GST fusion analyses of Fig. 1D activated both promoters with a potency similar to that of the wild-type construct (D22-A and ED30/31-AA in both Fig. 2A and B). However, mutations that were deficient in binding CBP were also defective for promoter stimulation (L26-A and DI28/29-AA). Similar results were obtained when the isolated activation domains (wild-type SREBP-1a and the mutants diagrammed in Fig. 1C) were fused to Gal4 and analyzed for activation of a GAL4-responsive reporter (data not shown).

FIG. 2.

Mutations that establish a correlation between SREBP-1a activation and coactivator interaction. SREBP-1a aa 1 to 490 (the mature protein) were inserted into 2XFLAGpCDNA3.1+, and the mutations diagrammed in Fig. 1C were introduced into this context. These constructs were cotransfected into 293T cells in increasing amounts (shown on the abscissa) along with luciferase reporter constructs containing either the HMG CoA synthase (A) or the LDL receptor (LDLR) promoter (B) and a control CMV-β-galactosidase plasmid that was used to normalize transfection efficiency. Fold activation was calculated as described in Materials and Methods. The inset in panel B shows an immunoblot analysis done with an anti-FLAG antibody to compare protein expression for the indicated constructs. Protein concentrations were normalized as described in Materials and Methods. (C) Evaluation of the GST fusion proteins analyzed in Fig. 1D for binding to in vitro transcribed-translated, ³⁵S-labeled peptide regions of either the KIX or the C/H1 domain of CBP or the entire DRIP150 coding sequence. Lane 1 is 10% of the input ³⁵S-labeled extracts used for each binding reaction.

Previous studies have indicated that SREBP-1a interacts with two different regions of CBP. In one study, an interaction between SREBP-1a and the KIX domain of CBP was identified (30) whereas another report showed that in addition to the KIX domain, the amino-terminal Zn finger C/H1 domain also interacts with SREBP-1a (12). To analyze the interactions between these two isolated domains of CBP with SREBP-1a and to determine which interaction was providing the binding pattern we established with full-length CBP in Fig. 1D, we evaluated interactions between the GST-SREBP fusions and the isolated CBP domains that were prepared by in vitro transcription-translation. The binding patterns for both CBP domains are shown in Fig. 2C. All four of the mutations analyzed for transcriptional activity as described above were deficient in binding of the isolated KIX domain of CBP, whereas the binding specificity with the C/H1 domain followed both the pattern observed for binding of full-length CBP from nuclear extract and the activation patterns observed in Fig. 2A and B. These results suggest that the C/H1 domain of CBP is responsible for the interaction with full-length CBP and that this is likely to be the critical interaction for SREBP-dependent gene activation.

As mentioned in the Introduction, the activation domain of SREBP-1a also interacts with the multisubunit mediator complex (31). In order to determine which subunit(s) of the mediator interacts with SREBP and to evaluate similarities and differences in how the mediator and CBP interact with SREBP-1a, we tested several mediator subunits for interaction with SREBP-1a (36). We independently expressed subunits referred to as DRIP77, -92, -100, -130, -150, -205, and -240 by in vitro transcription-translation and evaluated each one for interaction with a GST fusion of the SREBP-1a activation domain. Of the subunits tested, DRIP130 (data not shown) and DRIP150 (Fig. 2C) both interacted specifically with SREBP-1a. Interaction with the DRIP150 subunit by the SREBP-1a activation domain mutants was consistent with the binding pattern for the C/H1 domain of CBP and endogenous full-length CBP.

To further analyze the significance of the SREBP-KIX domain interaction, we evaluated SREBP activation in MEFs derived from wild-type mice or mice with targeted substitution mutations in the KIX domain of either CBP or p300 (WT, CBP^K/K, or p300^K/K). The three substitution mutations alter hydrophobic residues in the KIX domain (Tyr650Ala, Ala654Gln, and Tyr658Ala) and decrease binding to the α-helical activation domains of CREB and c-Myb. As a result, transcriptional activation by either CREB or c-Myb decreased significantly but activation by activators that do not interact with KIX was unaffected (23) (Fig. 3B). Sterol-dependent activation of the SREBP-responsive HMG CoA synthase promoter was 8.8-fold in the wild-type cells and was at least as robust in cells derived from the two mutants (Fig. 3A). We also directly evaluated binding of SREBP-1a to the CBP expressed in the wild-type and CBP^K/K mutant cells by a GST interaction assay similar to that used in the experiment whose results are shown in Fig. 1. Here, CBP from both cell lines interacted equivalently with SREBP-1a but once again SREBP-1c failed to interact with CBP derived from either cell line (Fig. 3C). These studies strongly suggest that under the conditions evaluated, the SREBP-KIX interaction is not required for SREBP-dependent activation.

FIG. 3.

The KIX domain of CBP is dispensable for SREBP-mediated gene expression and binding. (A) Primary wild-type (WT), CBP^K/K, and p300^K/K MEFs were transfected with the HMG CoA synthase promoter luciferase reporter and subsequently refed with DFSM minus sterols to induce endogenous SREBP expression (solid bars) or with DSFM plus sterols to suppress endogenous SREBP expression (dotted bars). Luciferase activities were normalized to β-galactosidase activity as described in Materials and Methods. Values indicate fold induction between individual induced and suppressed sample sets. (B) The MEFs from panel A were transfected with a constant amount of 5XGAL4UAS-Luc reporter and either the GAL4 DBD only or the GAL4 DBD fused to the VP16 activation domain (GAL4VP16). Luciferase values were normalized as in panel A. (C) GST fusions of mature SREBP-1a and -1c were immobilized on glutathione-agarose and incubated with nuclear extracts from immortalized wild-type and CBP^K/K MEFs. Interacting proteins were resolved by SDS-PAGE and detected by immunoblotting with an antibody to CBP or GST as indicated.

To this point, we have looked at coactivator interactions with SREBP-1a, focusing on residues that are conserved between SREBP-1a and SREBP-2 and are mostly absent from SREBP-1c. The data indicate that the conserved residues are important for coactivator interaction and are consistent with a model in which SREBP-1c is a weak activator because it lacks these conserved amino acids. However, SREBP-1c has a distinct amino terminus and neither the role of this unique peptide domain nor a systematic evaluation of how SREBP-1c interacts with coactivators has been reported.

To evaluate how mutations in the SREBP-1c activation domain influence its function, we fused the activation domains of both SREBP-1a and -1c to the Gal4 DBD and evaluated the activation of a Gal4-responsive reporter plasmid. In this context, there was still a dramatic difference in activation mediated by SREBP-1a relative to that mediated by SREBP-1c (Fig. 4A). Thus, the difference in activation mediated by SREBP-1a and -1c is retained in the heterologous Gal4 fusion context. This further supports the idea that the sole differences in activation by SREBP-1a and -1c are mediated by their distinct amino termini. Additionally, the level of activation is robust even for SREBP-1c. Because the level of activation observed for full-length SREBP-1c is close to the background level, we reasoned that the Gal4 fusion context provided a higher starting level that would allow a more sensitive comparison of how point mutations might affect activation by SREBP-1c.

FIG. 4.

Transcriptional activation and coactivator interactions mediated by SREBP-1c. (A) Gal4 fusions of the first 100 aa of wild-type SREBP-1a or the corresponding 75 aa of SREBP-1c were transfected into 293T cells in increasing amounts along with a constant amount of the 5XGAL4UAS-Luc reporter. Luciferase activity was normalized to β-galactosidase activity, and activation was calculated as described in Materials and Methods. (B) Mutations in unique exon 1 of SREBP-1c (illustrated in Fig. 1C) and a complete deletion of exon 1 were introduced into the Gal4-SREBP-1c hybrid construct used in panel A. The observed activation for each construct was compared as for panel A. Protein expression for each Gal4-SREBP-1 fusion was compared by immunoblotting with an antibody to Gal4 as described in Materials and Methods. (C) GST versions of the Gal4 fusions shown in panels A and B were expressed and purified from E. coli and used in GST interaction assays with the in vitro-transcribed-translated, ³⁵S-labeled CBP KIX and C/H1 domains or the full-length DRIP150 protein as described in the legend to Fig. 2. (D) SREBP-1c aa 1 to 466 (the mature protein) or a mutant version lacking exon 1 (Δ2-5) was inserted into 2XFLAGpCDNA3.1+ and transfected (30 ng/well) into CV-1 cells with a luciferase reporter construct containing the HMG CoA synthase promoter luciferase reporter construct. CMV-DRIP150 (100 ng/well) was included as indicated. Luciferase values were normalized to protein concentrations (see Materials and Methods). Protein expression was determined by immunoblotting with an antibody to the FLAG epitope.

We introduced mutations at all four unique SREBP-1c positions individually and also deleted the entire first exon (Fig. 4B and data not shown). When the resulting Gal4 fusions were evaluated for promoter activation, mutations that singly altered aa 2 to 4 did not impair activation (M2 in Fig. 4B; results for M3 and M4 are not shown). However, mutation M5 (F5-A) and deletion of the entire first exon resulted in proteins that were severely crippled for gene activation in this assay (Fig. 4B). All of these Gal4 fusion constructs were expressed at similar levels as measured by an immunoblotting experiment (Fig. 4B, inset). Therefore, the unique amino-terminal exon significantly contributes to promoter activation by SREBP-1c and the phenylalanine at position 5 is specifically critical.

To evaluate how these mutations affect SREBP-1c coactivator interactions, we performed interaction assays with GST chimeras of the same SREBP-1c peptides evaluated as Gal4 hybrid constructs with the in vitro-translated CBP domains and DRIP150 in the experiment whose results are shown in Fig. 2. Consistent with the results in Fig. 2, SREBP-1a interacted more efficiently with the two isolated CBP domains than did SREBP-1c (Fig. 4C, lanes 3 and 4). The same was true for DRIP150. In order to evaluate the low level of binding of DRIP150 to SREBP-1c, we added spermine to the binding reaction mixture. This was shown previously to enhance the in vitro interaction between mediator subunits and some nuclear receptors (27). In our experiments, spermine enhanced the SREBP-1-DRIP150 interaction but had no effect on the CBP interactions (data not shown). When the two SREBP-1c point mutations were analyzed for coactivator interaction, M2 was as effective as wild-type SREBP-1c in binding these coactivators. However, M5 failed to interact with any of the coactivator peptides. The same was true for the mutant with the entire SREBP-1c unique domain deleted (data not shown). Thus, the patterns for interaction by SREBP-1c and its mutant derivatives with coactivators are similar to the relative patterns for promoter activation. We also analyzed whether addition of an expression vector for DRIP150 would stimulate activation by mature full-length SREBP-1c in a transient transfection assay. Moderate increases in activity from an HMG CoA synthase promoter reporter were observed when a CMV-driven DRIP150 expression construct was coexpressed with SREBP-1c (Fig. 4D). Similar to the Gal4 fusion results, deletion of the unique SREBP-1c exon in the context of full-length SREBP-1c significantly decreased activation which was not stimulated by coexpression of DRIP150 (Δ2-5 in Fig. 4D). When put together, these data suggest that the weak activation by SREBP-1c is mediated through a comparably weak interaction with the same coactivators used efficiently by SREBP-1a. They also suggest that SREBP-1c's unique exon does contribute to coactivator binding and activation.

To further evaluate the SREBP-DRIP150 interaction, we made truncation mutations of DRIP150 and evaluated the interaction with SREBP-1a in a coimmunoprecipitation assay (Fig. 5A). c-Myc-tagged SREBP-1a was cotransfected with a construct expressing aa 1 to 824 of DRIP150 containing an amino-terminal FLAG epitope. Lysates from the transfected cells were immunoprecipitated with antibodies to the c-Myc epitope and then evaluated by immunoblotting for coprecipitation of the FLAG-tagged DRIP150 protein. Only in the reaction mixture in which both expression constructs were cotransfected was FLAG protein detected in the eluates from the immunoprecipitation (Fig. 5A, top). Immunoblotting with an anti-c-Myc antibody was used to verify that c-Myc-tagged SREBP was indeed expressed and efficiently immunoprecipitated (Fig. 5A, bottom). Next, we further localized the region of DRIP150 that interacts with SREBP-1a with additional truncation mutations in a GST interaction assay. As shown in Fig. 5B, residues 500 to 824 of DRIP150 interacted with SREBP-1a as efficiently as did the full-length protein.

FIG. 5.

Direct interaction between SREBP-1a and DRIP150 in cultured cells. (A) 293T cells were transfected with different combinations of 2XFLAGDRIP150 aa 1 to 824 and 2XMycSREBP-1a (either one individually or the two together) and processed into whole-cell lysates as described in Materials and Methods. Lysates (2 mg of total protein) were bound to Sepharose beads coupled with an antibody to the c-Myc epitope. Bound proteins were eluted, resolved by SDS-PAGE, and visualized by immunoblotting with antibodies directed to the FLAG or c-Myc epitope tag as indicated. Lanes 1 to 3 represent 1% (anti-c-Myc) and 0.2% (anti-FLAG) of the total cell lysates used in the immunoprecipitation (IP) reaction. Eluate lanes are 10% (anti-c-Myc) and 90% (anti-FLAG) of the total eluted volumes. (B) ³⁵S-labeled full-length DRIP150 or aa 500 to 824 was prepared by in vitro transcription-translation and tested for interaction with either GST protein alone or GST fused to the first 100 aa of wild-type (WT) SREBP-1a. Reactions were performed and visualized as described in Fig. 2C. Two different amounts (5 and 25 μl) of the in vitro-prepared DRIP150 protein extracts were used as indicated.

To further evaluate the importance and specificity of DRIP150 in SREBP-mediated gene activation in a cellular context, we independently expressed DRIP150 or several other mediator subunits in cells and evaluated their abilities to influence activation mediated by endogenously expressed SREBPs (Fig. 6). In this experiment, we transfected an HMG CoA synthase promoter reporter construct into cells and cultured them in the presence and absence of sterols such that endogenously expressed SREBPs were processed and targeted to the nuclei when the cells were cultured in the absence of an exogenous source of cholesterol (10, 11). As a result, the transfected promoter was stimulated significantly relative to that in parallel dishes of transfected cells that were cultured in the presence of regulatory sterols, where the nuclear accumulation of SREBPs was prevented. This regulation is shown by a comparison of lanes 1 and 2 in the top parts of Fig. 6A to D, where the luciferase activity was much higher in lane 2 when the cells were cholesterol depleted.

FIG. 6.

Cotransfection of DRIP150 specifically enhances SREBP-mediated gene activation. 293T cells were transfected with the HMG CoA synthase promoter luciferase reporter along with CMV-driven expression constructs for the individual DRIP subunits in increasing amounts (0, 0.01, 0.1, and 1 μg) as indicated in panels A to D. Transfected cells were refed with DFSM minus sterols to induce endogenous SREBP expression (solid bars) or with DSFM plus sterols to suppress endogenous SREBP expression (dotted bars). The graph at the top of each panel displays the effects of DRIP on the HMG CoA synthase luciferase reporter in the presence or absence of endogenously expressed SREBPs. Luciferase activities were normalized to protein concentrations (see Materials and Methods). The graph at the bottom of each panel illustrates the effects of independent DRIP subunit cotransfection on the activity of the CMV-β-galactosidase (β-gal) construct. RLU, relative light units.

When increasing amounts of an expression vector for full-length DRIP150 were included on top of the sterol depletion, fourfold stimulation of the HMG CoA synthase promoter was observed (Fig. 6A, lanes 5 and 6). When higher levels of DRIP150 are cotransfected, the activity of the promoter was repressed. These results are consistent with the lower level of DRIP150 stimulating SREBP-dependent activation and a higher level interfering with complex binding and thereby squelching activation. A similar conclusion was suggested in a previous study in which DRIP150 was shown to be important for STAT 2 activation (24). Importantly, these effects are specific as expression of a transfected CMV promoter-driven β-galactosidase expression construct was not significantly affected (Fig. 6A, bottom). Previous studies have shown similar results when other transcription factors were analyzed and the key interacting subunit of the mediator complex was also exogenously expressed (23). A similar transfection protocol for DRIP205, -240, and -77 resulted in only minimal effects on SREBP-dependent activation and CMV-β-galactosidase activity (Fig. 6B to D).

To provide direct evidence that DRIP150 protein is recruited to an SREBP target gene in cells in a sterol-responsive manner, we used the ChIP technique with a line of HeLa cells that stably express FLAG epitope-tagged DRIP150. These cells were cultured in the presence or absence of regulatory sterols for 24 h, cross-linked with formaldehyde, and then subjected to a ChIP analysis of DRIP150 binding to the endogenous sterol-regulated HMG CoA reductase promoter (Fig. 7). The results in Fig. 7A with total input DNA established that the PCR conditions used provided signals within the linear range of the amplification protocol. Samples from cells cultured in the presence or absence of sterols were analyzed in triplicate and show a significant increase in DRIP150 binding to the HMG CoA reductase promoter by the sterol depletion protocol (Fig. 7B). Control precipitations showed that recruitment was antibody dependent, while immunoblotting results indicated that there were equivalent levels of total DRIP150 protein in both cell extracts. An immunoblot also showed that SREBP-1 was induced properly by sterol deprivation and that inclusion of a proteosome inhibitor increased this response (Fig. 7B and C). These results provide direct evidence that sterol depletion in cells results in increased association of DRIP150 at an endogenous SREBP target promoter in cellular chromatin.

DISCUSSION

The present studies were designed to further evaluate the functional significance of different SREBP-CBP interactions in transcriptional activation, to identify the mediator subunit that directly interacts with SREBP-1, and to compare the patterns of coactivator interactions mediated by SREBP-1a and -1c in order to determine if the weak transcriptional activity of SREBP-1c could be explained by a corresponding weak interaction with CBP and/or the mediator complex. An earlier study compared the SREBP-1a activation domain with the KID of CREB since both interact with the KIX domain of CBP (5). Phosphorylation of the KID of CREB stimulated the interaction with KIX, whereas the SREBP-1a-KIX interaction is apparently constitutive. In order to explore the mechanism of the inducible versus constitutive interactions, the two-helix-containing KID was aligned with a region of predicted functional similarity within SREBP-1a and corresponding amino acids were exchanged between the two proteins in an attempt to convert CREB from kinase inducible to a constitutive KIX interactor. Transfer of the DIEDML peptide from SREBP-1a converted CREB into a relatively potent activator in the absence of kinase stimulation. However, these studies were focused on CREB, and no activation studies were performed with the mutated SREBPs. Also, the binding studies for SREBP were performed with full-length CBP in crude cell extracts and the pattern of interaction with the mutant SREBPs was presumed to be through the KIX domain. As mentioned above, both the KIX and C/H1 domains interact with the SREBP-1a activation domain so it was not clear how the mutations in this study specifically affected interactions between the two separate CBP domains and SREBP-1a.

In our initial studies, we concentrated on similarities and differences between the two SREBP-1 proteins and SREBP-2 and first evaluated the significance of the predicted α-helical activation domain that is common to both SREBP-1a and SREBP-2 by making mutations in the conserved residues. We also isolated the C/H1 and KIX domains and compared their binding to the SREBP-1a activation domain and derived mutants with binding of endogenous CBP. Mutation of the single conserved D-22 or the doublet E-D at positions 30 and 31 of SREBP-1a did not have a significant effect on binding to full-length CBP or the isolated C/H1 domain, but both mutations significantly decreased binding to the isolated KIX domain. Importantly, these two mutant proteins still activated SREBP target promoters to levels similar to that of the wild-type protein. In contrast, mutation of either conserved L26-A or the doublet D-I at positions 28 and 29 abolished activation (Fig. 2A and B) and decreased interaction with full-length CBP (Fig. 1D), as well as with both the isolated KIX and C/H1 domains (Fig. 2C). Taken together, these observations suggest that SREBP-1a and -2 activate transcription by interacting with the C/H1 domain of CBP.

We also analyzed SREBP activation and interaction with CBP in MEFs from wild-type and mutant mice with targeted amino acid substitutions in the KIX domain of either CBP or p300 (Fig. 3). These studies also suggest that the KIX domain is not required for SREBP binding or transcriptional activation. In contrast, activation mediated by the CREB and c-Myb transcription factors, both of which interact with the KIX domains of CBP and p300, was significantly decreased in the mutant MEFs (23).

The original studies that identified CBP as a coactivator for SREBP transcription (33) localized an interacting domain to the KIX domain by in vitro binding to the SREBP-1a activation domain. When a peptide containing the KIX domain was added to a reconstituted in vitro transcription system, it inhibited synergistic activation mediated by SREBP-1a with Sp1, presumably by a dominant negative mechanism. Another report identified aa 1 to 451 of CBP as an SREBP-1a-interacting peptide that also inhibited SREBP-dependent gene activation when overexpressed in transfected cells (12). This region contains the C/H1 zinc finger domain (aa 345 to 439), which also interacts with several other proteins (16). While the CBP fragment containing the KIX domain also interacted with SREBP-1a in this second study, it did not inhibit SREBP-dependent gene activation although it was not clear if this CBP fragment was expressed in the system under investigation.

Conclusions from these studies were based on interpretations of dominant negative inhibition experiments and were not completely concordant. While potentially very powerful, dominant negative approaches need to be interpreted cautiously because any peptide fragment that binds to the SREBP-1a activation domain and blocks an important coactivator interaction would score positive in this assay. It is important to reinterpret these studies in the light of our new data that show that the DRIP150 subunit and both the C/H1 and KIX domains interact with overlapping regions of SREBP-1a. Thus, when overexpressed, peptides from both CBP domains would inhibit interaction with the mediator complex and both domains of CBP. Taken together, all of the available data indicate that SREBP-1 interacts with CBP through its C/H1 domain to activate transcription and that when overexpressed, the KIX domain alone can interfere with this interaction. It remains possible that the interaction between SREBP-1a and the KIX domain is important for activation of select promoters where the combination of SREBPs and other required transcription factors makes it a preferred target interaction domain.

The SREBP-1a activation domain also interacts with the multisubunit mediator complex (31), and our studies are the first to demonstrate that the mediator complex is recruited to an endogenous SREBP promoter in a sterol-sensitive manner. We also provide in vitro evidence that SREBP-1 makes a key interaction with the 150-kDa mediator subunit (referred to here as DRIP150) and that SREBP-dependent transcriptional activity is upregulated when DRIP150 is coexpressed in transfected cells. DRIP150 is a highly conserved mediator subunit that is present in all metazoans and in yeast (28). The yeast subunit, Rgr1, was first identified as a protein encoded by a gene required for glucose repression in Saccharomyces cerevisiae, and a null mutation resulted in the inability to accumulate carbohydrates (40). Rgr1 is part of a subcomplex that contains several other mediator polypeptides, and an RNA interference knockdown approach targeting DRIP150 or RGR1 in Caenorhabditis elegans resulted in decreased phosphorylation of select residues of the RNA polymerase II carboxy-terminal domain and in a broad range of developmental defects consistent with a global role in gene expression (45).

Other transcriptional activators that interact with the mediator complex through the DRIP150 subunit include the glucocorticoid receptor (GR) (19) and STAT2 (24). The AF-1 domain of the GR interacts through a region at the carboxyl terminus of DRIP150 (aa 1360 to 1454). This interaction is required for gene activation by the GR but is not directly regulated by ligand binding (19). STAT2 interacts with an amino-terminal domain of DRIP150 (aa 188 to 566) that is important for activity of STAT 2 in the interferon response (24). We identified the region between aa 500 and 824 in DRIP150 as the SREBP-interacting domain (Fig. 5) that in part overlaps the fragment that interacts with STAT2. Thus, DRIP150 is a multifunctional protein capable of binding distinct transcriptional regulators through separate interaction domains.

Our studies indicate that CBP and mediator interact with an overlapping region of SREBP-1a, suggesting that the interactions are mutually exclusive. This is not an uncommon observation, as the C-terminal activation domain of STAT2 that binds DRIP150 also interacts with the CBP C/H1 domain (34). A similar overlap in the interaction regions exists for nuclear receptors and different coactivators (39). A careful time course analysis of coactivator recruitment by the estrogen receptor following ligand addition indicated that the coactivators CBP and mediator bind and dissociate from the estrogen receptor in cyclic patterns that are out of phase with each other (4). It is possible that the same is true for coactivator recruitment by SREBP-1a. However, the synchronous and precise timing achieved by addition of nuclear receptor ligands to cell culture medium would not be possible to replicate for SREBP activation, which is a multistep process initiated by intracellular sterol depletion and results in an ill-defined cell-autonomous physiological response that would not likely occur simultaneously in every cell in a population.

A significant portion of the predicted α helix encompassing the SREBP-1a activation domain is not present in SREBP-1c. Instead, SREBP-1c has an amino-terminal domain with four unique residues. As reported here, the weak activation mediated by SREBP-1c relative to that mediated by SREBP-1a is maintained when the individual activation domains are fused to the Gal4 DBD. This is strong evidence that the differences in activation by the two different SREBP-1 isoforms are solely attributable to their unique respective amino termini.

The unique amino-terminal domain of SREBP-1c does contribute to its transcriptional activity (Fig. 4). As heterologous fusions with Gal4, individual mutations at aa 2 to 4 did not significantly reduce activation whereas an F5-A mutation or deletion of all four SREBP-1c-specific residues completely abolished activation by the Gal4 fusion construct (Fig. 4B). Deletion of the unique exon in full-length SREBP-1c also severely decreased activation (Fig. 4D). The results of GST binding studies with the SREBP-1c variants showed that mutant proteins that still activated transcription weakly interacted weakly with CBP and DRIP150 and that mutations that decreased binding also decreased activation. Because the weak interactions between SREBP-1c and either CBP or DRIP150 were specifically disrupted by mutations that abolish promoter stimulation, these results are consistent with a model in which SREBP-1c weakly activates transcription through weak contacts with the same coactivators that interact with SREBP-1a.

The mature SREBPs turn over rapidly through the ubiquitin-proteasome system (18). A recent report suggested that the level of gene activation by SREBP-1a and SREBP-2 is reciprocally related to their stability (51). These findings predict that SREBP-1c, a weak activator, should accumulate to levels significantly higher than those of SREBP-1a. It also predicts that mutations in the SREBP-1a activation domain that decrease its activation potential should in turn increase accumulation. In contrast, all of the mutant proteins analyzed in our study accumulated to similar levels regardless of their potency of activation. Two earlier studies also examined SREBP-1 expression and transcriptional activation. In one, SREBP-1a and -1c expression and activity were compared over a wide range of absolute protein expression levels (50), and in the other, the amino-terminal activation domain of SREBP-1a was defined through deletion analyses, comparing protein expression and transcriptional activity of the resulting mutant proteins (43). In both of these studies, SREBP-1a, SREBP-1c, and all of the mutant SREBP-1 proteins accumulated to similar levels, regardless of their potency in gene activation. Importantly, our results agree with these earlier reports and suggest that the relationship between SREBP stability and transcriptional potency may be more complex than a simple inverse relationship. This topic warrants further investigation.

There is only one SREBP gene and only one identified protein isoform in Drosophila melanogaster, and it resembles SREBP-1a (38). The reasons for having both SREBP-1a and -1c in mammals are not clear. Because SREBP-1a both interacts with coactivators and activates gene expression more robustly than SREBP-1c, it is likely that SREBP-1a would be expressed under conditions in which increased lipid synthesis is required and SREBP-1c would be preferentially expressed in tissues and physiological situations in which relatively low but regulated levels of lipid synthesis are important. The available data support this hypothesis, as SREBP-1a is the predominant isoform in continuously growing cultured cells and in the spleen, where actively turning over cells of the immune system are located. Since SREBP-1c is the predominant isoform in all of the other tissues examined (50), it would make sense that when moderate target gene expression is required, a less robust activator would orchestrate the downstream response. It is also possible that an unidentified cell-specific coactivator also plays an important role in differential gene activation by SREBP-1a and -1c.