Ras Signaling and Transcriptional Synergy at a Flexible Ets-1/Pit-1 Composite DNA Element Is Defined by the Assembly of Selective Activation Domains*

Pit-1 and Ets-1 binding to a composite element synergistically activates and targets Ras-mitogen-activated protein kinase signaling to the rat prolactin promoter. These transcriptional responses appear to depend on three molecular features: organization of the Ets-1/Pit-1 composite element, physical interaction of these two factors via the Pit-1 homeodomain (amino acids 199–291) and the Ets-1 regulatory III domain (amino acids 190–257), and assembly of their transcriptional activation domains (TADs). Here we show that the organization of the Ets-1/Pit-1 composite element tolerates significant flexibility with regard to Ras stimulation and synergy. Specifically, the putative monomeric Pit-1 binding site can be substituted with bona fide binding sites for either a Pit-1 monomer or dimer, and these sites tolerated a separation of 28 bp. Additionally, we show that the physical interaction of Ets-1 and Pit-1 is not required for Ras responsiveness or synergy because block mutations of the Pit-1 interaction surface in Ets-1, which reduced Ets-1/Pit-1 binding in vitro, did not significantly affect Ets-1 stimulation of Ras responsiveness or synergy. We also show differential use of distinct TAD subtypes and Pit-1 TAD subregions to mediate either synergy or Ras responsiveness. Specifically, TADs from Gal4, VP16, or Ets-2 regulatory III domain linked to Ets-1 DNA binding domain constructs restored synergy to these TAD/Ets-1 DNA binding domain fusions. Conversely, deletion of the defined Pit-1 TAD (amino acids 2–80) retained synergy, but not Ras responsiveness. Consequently, we further defined the Pit-1 amino-terminal TAD into region 1 (R1, amino acids 2–45) and region 2 (R2, amino acids 46–80). R1 appears to regulate basal and synergistic responses, whereas the Ras response was mapped to R2. In summary, Ras responsiveness and Pit-1/Ets-1 synergy are mediated through the assembly of distinct TADs at a flexible composite element, indicating that different mechanisms underlie these two transcriptional responses and that the Pit-1 R2 subregion represents a novel, tissue-specific Ras-responsive TAD.

In precisely this manner, an Ets/Pit-1 combination, acting via the Ets binding site (EBS)/footprint IV (FPIV) composite element spanning positions Ϫ217 to Ϫ190, is critical for both basal lactotroph-specific expression of the rPRL gene and for the targeting of the Ras-MAP kinase pathway to the prolactin promoter (3,10). Several lines of evidence indicate that both Ets-1 and Pit-1 factors are required for the synergistic activation and Ras responsiveness. For example, transient transfection of pituitary-derived GH4T2 somatolactotrope cells with a dominant-negative Ets factor encoding only the Ets DNA binding domain (DBD), or with Pit-1␤ (an alternatively spliced isoform that functions as a dominant-negative effector in pituitary cells), results in diminished activity of the rPRL promoter in an effector-dependent manner (3). Similarly, transient transfection of rPRL promoter constructs containing site-specific mutations of either the EBS or FPIV sites within the composite element results in a loss of Ras responsiveness (10,11). In non-pituitary HeLa cells, it has been shown that although individual Ets-1 or Pit-1 effectors strongly activate the rPRL promoter, the combination results in a response that is 6-fold greater than the sum of the individual responses (i.e. 6-fold synergy) (3). In addition, Ets-1 and Pit-1 have been shown to bind physically to each other in dilute solution and in the absence of DNA by utilizing the Ets-1 regulatory III (RIII) transcriptional activation domain (TAD) and Pit-1 homeodomain (3,12,13). Finally, Thr-82 of chicken p68 c-Ets-1 was identified as a MAP kinase phosphorylation site, and sitespecific mutation of Thr-82 to alanine resulted in loss of the Ets-1-mediated enhancement of the Ras response (14). Taken together, these results support a two-step model of Ets-1/Pit-1 activation of the rPRL promoter in which binding of Ets-1/Pit-1 to the composite element and to each other regulates basal promoter activity, and Ras activation occurs in part as a response to phosphorylation of Ets-1 at Thr-82.
Structure-function analysis of p68 Ets-1 has highlighted a number of features. Specifically, the TAD is composed of three regulatory domains, RI-RIII. RI and RIII function as positive TADs, whereas RII functions as a domain that negatively regulates the function of RI and positively regulates RIII. Overlapping the RII TAD is the pointed domain which contains the MAP kinase phosphorylation site, Thr-82. The latter is highly conserved in the Ets subfamily, comprising Ets-1, Ets-2, and pointed P1 and P2 (15). In the Ets subfamily, RI and RIII are unique to either Ets-1 or Ets-2. Amino acids 180 -300 within RIII have only 6% similarity between Ets-1 and Ets-2 compared with ϳ90% similarity for the DBD and 66% conservation for RII. Finally, the ETS DBD, located near the carboxyl terminus of Ets-1, is flanked by two domains that inhibit DNA binding, In1 and In2. The DBD is a highly conserved 84 -90amino acid domain referred to as the ETS domain (16,17). The ETS domain, which serves to define the family, utilizes a winged helix-turn-helix structural motif to bind to Ets binding sites. The core sequence of the Ets binding sites consists of the sequence 5Ј-GGA(A/T)-3Ј. In addition, flanking DNA sequences on either side of this core binding element may contribute to specific Ets factor binding, such that the actual binding site may consist of ϳ9 bp (15). The composite EBS/FPIV binding site located at Ϫ217 to Ϫ190 consists of a core GGAA site on the antisense strand separated by 8 bp from the core binding site for Pit-1 (see Fig. 1). Typically, monomeric Ets factors bind to DNA weakly and require a vicinally bound factor to stabilize Ets factor-DNA interactions (18). Consequently, Ets factors and their partners are often recruited to bipartite binding sites on DNA, forming ternary complexes and resulting in synergistic transcriptional responses. The spacing and orientation of these bipartite binding sites do not appear to be critical determinants for the binding of these ternary complexes. For example, ternary complexes of SRF and the Ets factors Elk-1 and SAP-1 can occur even when the respective binding sites are inverted with respect to each other or moved apart over two helical turns of DNA (19). Rather, DNA binding appears to be modulated by the flanking autoinhibitory sequences, In1 and In2, which can be regulated by specific protein partnerships and further modulated by protein phosphorylation (15, 18, 20 -23). Indeed, the most important mechanism for achieving target gene specificity is cofactor-induced alterations of Ets protein interactions. Usually, Ets proteins interact with partners through the EtsDBD (15,18,22,23). However, corebinding factor and Pit-1 interact with Ets-1 via domains distinct from the DBD, with core-binding factor binding to the In1 autoinhibitory domain and Pit-1 binding to the RIII TAD (3,24). Further analysis has refined the Pit-1 interaction face of Ets-1 to a 67-amino acid subdomain (amino acids 190 -257) within this Ets-1 isoform-specific region (3,12).
Conversely, POU homeodomain transcription factors utilize a bipartite DBD that can partially encircle the DNA. The DNA element thus dictates the configurations of subdomains and subsequent recruitment of specific coregulators to control transcription (25). Specifically, the ability of Pit-1 to bind to DNA as either a monomer or a dimer dictates the domains that it uses to synergize with heterologous transcription factors (26). Similarly, the spacing of the contact points for the POU-specific domain and POU homeodomain is sufficient to transform Pit-1 from a trans-activating factor to a repressor (27). In addition to its role in DNA binding, the homeodomain of Pit-1 is necessary and sufficient for physical interaction with Ets-1. Although it is unclear what precise role the physical interaction of Ets-1/Pit-1 plays in the synergistic activation of the rPRL promoter, it is intriguing to note that the ␤-domain of the Pit-1␤ isoform also interacts with this Ets-1-specific region (13).
In this study we examined the role played by spacing and composition of the combinatorial EBS/FPIV element or the physical interaction of Ets and Pit-1 in Ras responsiveness and synergistic activation of the rPRL promoter. We show that the Ets-1/Pit-1 composite element represents a highly flexible site through which the assembly of select transcriptional activation domains stimulates synergistic activation and Ras responsiveness of the rPRL promoter. Moreover, we show that the Ets-1/ Pit-1 synergism and Ras responses require different subdomains of the Pit-1 TAD, indicating that these two responses are mediated by distinct transcription control mechanisms. Taken together, these results provide novel insights into molecular mechanisms that specify transcription synergy and oncogenic Ras transcription responses.

Plasmid Construction
pA3-425PRL-Luc containing the proximal 425 bp of 5Ј-flanking sequence from the rPRL promoter in the pA3Luc reporter has been described previously (28). Mutations mEBS and mFPIV have been previously described (11). Mutations of pA3-425PRL-Luc including monomer, dimer, 12 bp, 16 bp, and 24 bp were generated by overlap extension PCR amplification from Ϫ425PRLpGEM7 using mutant internal primers and SP6 and T7 primers with dual rounds of amplification. The 12-bp and 16-bp insertions incorporate an SalI site into the space between the EBS and FPIV Pit-1 binding site. The 24-bp insertion incorporates SalI, SpeI, and NheI sites. The mutant primers are as follows: monomer S, GCA TTA AAA AAT GCA TAT CCT TCC; monomer  AS, GAT ATG CAT TTT TTA ATG CAA AAG G; dimer S, CTT TTG ATG  TAT ATA CAT AAA ATC C; dimer AS, GAT TTT ATG TAT ATA CAT  CAA AAG G; 12-bp S, TCC TTT TGT CGA CTG TAA TTA ATC AAA  ATC C; 12-bp AS, ACA GTC GAC AAA AGG AAA TGA GAG A; 16-bp  S, TGT CGA CGT CCT GTA ATT AAT CAA AAT CC; 16-bp AS, TAC  AGG ACG TCG ACA AAA GGA AAT GAG AGA; 24-bp S, TCG ACA  CTA GTG CTA GCT GTA ATT AAT CAA AAT CC; 24-bp AS, CTA GCA CTA GTG TCG ACA AAA GGA AAT GAG AGA. Mutant promoter 28 bp was generated by digesting the 24-bp insertion construct with SpeI, filling in the overhangs by Klenow treatment, and religating the vector. Mutant promoter 32 bp was generated by digesting 28 bp with SalI, filling in the overhangs by Klenow treatment, and religating the vector. Sequences were verified by sequencing.
The mutant Ets-1/2 chimera was generated using overlap extension PCR to replace amino acids 185-257 of Ets-1 with amino acids 178 -251 of c-Ets-2. The resulting PCR product was digested at internal BglII and AflIII restriction sites and ligated into a pSG5 Ets-1 vector in which the BglII site in the multiple cloning site was destroyed. This vector was partially digested with AflIII and completely digested with BglII. The primers used for PCR were as follows: the previously listed p68 Ets-1 401 S and p68 Ets-1 1054 AS; Ets1-2A, TAT TGA TCT GGG TAT GGT TTT GCC T; Ets1-2B, CCA TAC CCA GAT CAA TAT GTG GAG AG; Ets1-2C, GTC TGG ACT TTA GGG AAC AAT GAA AAG; Ets1-2D, TCC CTA AAG TCC AGA CGG ACT CC.

GST Pull-down Assays
GST Fusion Protein Preparation-Recombinant fusion proteins were prepared from bacterial extracts. Overnight cultures of Escherichia coli BL-21 (DE3)pLysS (Stratagene), transformed with pGex plasmids, were diluted 1:10 in fresh Luria broth supplemented with 100 g/ml ampicillin and grown at 30°C to an absorbance of 0.5 at 600 nm. The cultures were induced by addition of isopropyl-␤-D-thiogalactopyranoside to a final concentration of 1 mM. Cultures were grown for an additional 2 h at 30°C. The bacteria were harvested by centrifugation at 5,000 ϫ g for 10 min at 4°C and stored at Ϫ80°C until preparation. The bacterial pellets were resuspended in 5 ml of Bugbuster Reagent (Novagen)/g of bacterial pellet supplemented with a 1ϫ concentration of Complete protease inhibitor mixture (Roche Applied Science) and 25 units of benzonase/ml of reagent and incubated at room temperature with rocking for 20 min to allow for lysis. Alternatively, some preparations were resuspended in a 1/10 volume of phosphate-buffered saline (PBS) supplemented with a 1ϫ concentration of Complete protease inhibitor mixture and sonicated for three 10-s bursts at 4°C. Following sonication, Triton X-100 was added to a final concentration of 1%, and the bacterial pellets were incubated with rocking at room temperature for 30 min. If necessary, bacterial DNA was sheared with three additional bursts of sonication. Following lysis, cellular debris was removed by centrifugation at 16,000 ϫ g for 20 min at 4°C. The supernatant was transferred to a clean tube and bound to glutathione-Sepharose CL-4B (Amersham Biosciences) for 1 h at room temperature. The Sepharose was washed extensively with 1ϫ PBS supplemented with Complete protease inhibitors and 1 mM dithiothreitol. Ets-1 fusions were supplemented with 10 mM dithiothreitol. Protein concentration was measured by the Bio-Rad assay (Bio-Rad). Bound protein was analyzed for intactness and mass by SDS-PAGE in parallel with a known amount of bovine serum albumin and Gelcode Blue or Coomassie Blue staining.
In Vitro Binding Assays-35 S-Labeled proteins were synthesized and labeled with [ 35 S]methionine (PerkinElmer Life Sciences) using the TNT-coupled transcription-translation reticulocyte lysate system with T7 polymerase and supercoiled plasmids pSG5-Ets-1, BPV-1, BPV-2, BPV-3, Ets-2, Ets-1/2, VP16EtsDBD, and Gal4AD⌬5-6 according to the manufacturer's protocol (Promega, Madison, WI). Approximately equal amounts (5 g) of GST fusion proteins bound to 20 l of glutathione-Sepharose beads were suspended in a total volume of 0.5 ml of binding buffer (40 mM HEPES, pH 7.5, 100 mM NaCl, 5 mM MgCl 2 , 0.1 mM EDTA, 0.05% Nonidet P-40, 1 mM dithiothreitol, and 1ϫ Complete protease inhibitors). Ethidium bromide was added to a final concentration of 50 g/ml to eliminate binding of any contaminating DNA (31) and incubated for 30 min at 4°C on a rotator. 35 S-Labeled proteins (5 l of a 50-l synthesis reaction) were added to the GST fusion samples and incubated on a rotator at room temperature for 1 h. The Sepharose beads were pelleted by centrifugation at 2,500 ϫ g for 5 min. Aliquots of the supernatant were reserved to assess input, and the beads were washed four times with 1 ml of binding buffer containing 0.1% Triton X-100. The 35 S-labeled proteins were eluted from the beads by boiling in SDS-sample buffer and analyzed by SDS-PAGE and autoradiography. Bands were quantified using a Molecular Dynamics PhosphorImager with ImageQuant software.

Cell Culture
HeLa and GH4T2 cells were maintained in Dulbecco's modified Eagle medium (Invitrogen) supplemented with 15% horse serum and 2.5% fetal calf serum (Invitrogen). Cells were grown at 37°C in 5% CO 2 . The medium was changed 4 -16 h prior to transfection.

Transient Transfection
Electroporation-Cells were harvested with 1ϫ PBS with 3 mM EDTA and resuspended in culture medium. Aliquots of ϳ2-4 ϫ 10 6 cells in 200 l of medium were added to plasmid DNA as described in the figure legends and transfected by electroporation at 220 V and 500 microfarads using a Bio-Rad Gene Pulser with 4-mm gap cuvettes. Following transfection, cells were plated on 60-mm tissue culture plates in culture medium and incubated for 24 h. All electroporations included 0.3 g of CMV-␤-galactosidase (Clontech) as an internal control for transfection efficiency. Total DNA was kept constant, and nonspecific effects of viral promoters were controlled for by transfecting appropriate control vectors. Approximately 24 h following the electroporation, the cells were harvested, and luciferase and ␤-galactosidase activities were measured. The transfected cells were harvested in PBS containing 3 mM EDTA, pelleted by centrifugation, and lysed by three freeze-thaw cycles in 100 l of 100 mM potassium phosphate, pH 7.8 with 1 mM dithiothreitol. Cell debris was pelleted by centrifugation at 10,000 ϫ g for 10 min at 4°C, and aliquots of the supernatant were used in subsequent assays. Samples were assayed in duplicate for luciferase activity as described previously (28) using a Monolight 3010 luminometer (Analytical Luminescence Laboratories). ␤-Galactosidase activity was determined spectrophotometrically using the chromogenic substrate o-nitrophenyl-␤-D-galactopyranoside as described (28).
Lipid-mediated Transfection-HeLa cells were transiently transfected by lipid-mediated DNA transfer using Effectene (Qiagen) according to the manufacturer's instructions. Briefly, on day 1, HeLa or GH4T2 cells were harvested in 0.05% trypsin with 0.5 mM EDTA and plated on 96-well plates at a concentration of 20,000 or 45,000 cells/well in 100 l of medium, respectively. On day 2, plasmid DNA for transfection was prepared by mixing 0.8 l/well of Enhancer and 2.5 l/well Effectene reagent diluted in 30 l of EC buffer/well. Where indicated, DNA concentrations of the Pit-1 and Ets constructs were adjusted to give approximately equal levels of expression as determined previously (13). At 24 h following transfection, the cells were harvested and firefly luciferase and Renilla luciferase activity were determined on a Dynex microtiter plate luminometer using the Dual luciferase reporter assay system (Promega). Briefly, cells were washed once with 1ϫ PBS and lysed by adding 40 l of Passive lysis buffer (Promega) and incubating at room temperature with shaking for 15 min. The cell lysate was transferred to an opaque 96-well plate, and firefly luciferase and Renilla luciferase activity were measured for 10-s intervals following the addition of either 50 l of luciferase reagent or Stop-N-Glo reagent. Total DNA was held constant, and each well was transfected with 0.1 ng of pRLC-Renilla as an internal control for transfection efficiency. We controlled for nonspecific effects of viral promoters by transfecting appropriate control vectors.

No Preference Is Revealed for Pit-1 Monomer versus Dimer Binding Sites for Ras Stimulation and Synergy with Ets-1-
Deletion and mutational analysis of the rPRL promoter has identified the composite EBS/FPIV region as being critical for Ras responsiveness (10,11). This composite element consists of an EBS separated by 8 bp from a putative monomeric Pit-1 binding site (Fig. 1A). To define the role that the composition of these two sites plays in synergistic activation and Ras responsiveness of rPRL, we generated mutations that replace the WT FPIV Pit-1 binding site with the 1D monomer site derived from the Pit-1/ER composite enhancer element (monomer) (26) and a palindromic FPI site shown to bind DNA as a dimer by crystallography (1) (Fig. 1A). We tested these mutations in the context of pA3-425rPRL-Luc for Ras responsiveness in tran-sient transfections of GH4 cells. Block mutations of either EBS or FPIV were tested as positive controls for loss of function (10,11). Neither of the mutations significantly reduced the ability of the Ϫ425PRL promoter to respond to Ras stimulation as determined by Student's t test, p Ͻ 0.05 (Fig. 1B). However, in agreement with previously published work, block mutations in either the EBS or FPIV site significantly reduced the Ras response (11).
To characterize the synergistic response of each monomer/ dimer substitution construct in the HeLa reconstitution assay, we first tested the response of these constructs in the absence of either Pit-1 or Ets-1 (Fig. 1C, Vector), to Pit-1 alone (ϩPit-1), Ets-1 alone (ϩEts-1), and finally to Pit-1 plus Ets-1 (Both). Because the activity of the rPRL promoter is highly restricted to Pit-1-positive pituitary cells, transfection of each rPRL promoter-Luc reporter construct alone resulted in minimal activity, with relative light units (RLU; firefly luciferase divided by Renilla luciferase light units) in the range of 0.01 (Fig. 1C, Vector), as reported previously (3). However, coexpression of Pit-1 resulted in a 170-fold stimulation of the WT rPRL promoter, increasing the RLU from ϳ0.01 in the absence of Pit-1 to ϳ1.7 RLU in the presence of Pit-1 (Fig. 1C, ϩPit-1). The Pit-1-enhanced activity of the monomer, dimer, and mEBS re-porter constructs revealed a slight reduction in promoter activity compared with WT which was not statistically significant. However, the effects of Pit-1 on the mFPIV construct resulted in ϳ0.7 RLU, or about 70-fold stimulation, which was significantly reduced compared with the WT rPRL promoter, indicating that this Pit-1 binding site, one of three found in the Ϫ425PRL promoter, is required for optimal Pit-1 stimulation. Cotransfection with an Ets-1 expression vector resulted in ϳ30-fold stimulation of the WT and dimer rPRL promoters, increasing the RLU from ϳ0.01 in the absence of Ets-1 to ϳ0.3 RLU in the presence of Ets-1 (Fig. 1C, ϩEts-1). The monomer construct was about twice as responsive to Ets-1, whereas the mEBS and mFPIV promoters displayed somewhat reduced activities compared with WT (Fig. 1C, ϩEts-1). Finally, the combination of Pit-1 plus Ets-1 resulted in the expected strong synergistic, ϳ570-fold response of the WT promoter construct, increasing the RLU from ϳ0.01 in the absence of Pit-1/Ets-1 to ϳ5.7 RLU in the presence of Pit-1ϩEts-1 (Fig. 1C, ϩBoth). Moreover, the dimer and mFPIV promoter constructs maintained a synergistic response equivalent to that of WT, whereas the monomer and mEBS promoter constructs displayed modest increases in the synergistic response (Fig. 1C, ϩBoth). These HeLa reconstitution data reveal that the Ets-1/Pit-1 composite Separation of the Ets-1 and Pit-1 Sites by 32 Base Pairs Is Required to Reduce Ras and Synergistic Responses-We next asked how spacing of the EBS/FPIV binding sites affected Ras responsiveness in GH4 pituitary cells and the synergistic response in the HeLa reconstitution assay, by increasing the spacing between the sites from 8 bp to 12, 16, 24, 28, and 32 bp ( Fig. 2A). We found that increasing the spacing to 12 bp resulted in a significant increase in Ras responsiveness, from 7to 12-fold (Fig. 2B). Although responsiveness was still elevated at 16 bp, the increase was no longer significant. Subsequent increases in spacing resulted in a gradual loss in Ras responsiveness until separation of the sites by 32 bp resulted in a significant reduction in Ras responsiveness to 4-fold, similar to the loss in activity shown with block mutation of either the EBS or FPIV site (ϳ4-fold each). In the HeLa reconstitution assay, the WT and mutant promoters displayed very low basal activity in the 0.01 RLU range (Fig. 2C, Vector). Cotransfection with Pit-1 alone resulted in the expected strong ϳ115-fold activation of the WT promoter, from 0.01 to 1.15 RLU (ϩPit-1). The increased spacing between the Ets and Pit-1 binding sites resulted in a gradual decrease in the activation of the PRL promoter by Pit-1 alone, with the 12-, 16-, 24-, 28-, and 32-bp insertions resulting in ϳ95-, 80-, 70-, 50-, and 30-fold stimulation, respectively (Fig. 1C, ϩPit-1). In contrast, Ets-1 stimulated all of the spacing mutants to the same extent as the WT promoter (ϳ15-fold), as assessed by Student's t test (Fig. 2C, ϩEts-1). Finally, the synergistic stimulation by the combination of Pit-1 and p68 c-Ets-1 resulted in an apparent gradual decrease from ϳ600-fold for the WT promoter to ϳ200-fold for the 32-bp insertion (Fig. 2C, ϩBoth). These data reveal that the Ets-1/Pit-1 composite element allows modest spacing between the two sites for both the Ras and HeLa synergistic responses, yet there is a maximum spacing of 28 bp which is tolerated in these two functional assays.
Scanning Mutations of Ets-1 Amino Acids 190 -257 Reduce Pit-1 Binding but Not Ras or Synergy Responses-Previously, studies in our laboratory using GST pull-down analysis have shown that Pit-1 physically binds to Ets-1 in the absence of DNA. In Pit-1, this physical interaction has been mapped to amino acids 199 -291, which define the Pit-1 homeodomain. GST interaction assays and NMR studies have shown that both amino-and carboxyl-terminal contacts in the Pit-1 homeodomain are required for interaction with Ets-1 (12, 13). Carboxyl- and amino-terminal deletion analysis of Ets-1 has shown that the Pit-1 interaction domain maps to amino acids 190 -257 within the RIII TAD (3,12). To map further the domain in Ets-1 responsible for binding to the Pit-1 homeodomain, we replaced overlapping 26-amino acid segments of p68 c-Ets-1 in the region from 190 to 257 with amino acids 415-440 of the BPV L-1 capsid (including the AU1 epitope) to generate a series of three block mutations that "scan" across this 68-amino acid region (BPV-1-3, Fig. 3A) (32). GST pull-down analysis was used to assess binding of these constructs to the Pit-1 homeodomain. Fig. 3B shows that WT Ets-1 bound 17% of input, whereas mutant BPV-1 and BPV-2 proteins bound 8 -9% of input, a decrease of ϳ50%. Mutation of amino acids 232-257 (BPV-3) resulted in only 5% binding, or a 71% reduction compared with WT Ets-1 (Fig. 3B). These results suggest that there are multiple contact points for Pit-1 within amino acids 190 -257, with a major site located in the region mutated by BPV-3. Despite these reductions in the in vitro interaction assay, these mutations had minimal effects, if any, in the in vivo functional assays (Fig. 4). For example, in the absence of Ras stimulation, these BPV mutant Ets constructs resulted in an apparent slight increase in basal rPRL promoter activity compared with WT Ets-1; however, these increases were not statistically significant (Fig. 4A, black bars). With regard to the Ras response, WT Ets-1 enhanced the Ras response from ϳ7-fold in its absence to ϳ14-fold in its presence (Fig. 4A, white bars). Each BPV Ets-1 mutant enhanced the Ras response in a similar fashion, ranging from 12-to 14-fold and did not differ from the WT Ets-1 construct (Fig. 4A). The intrinsic transcriptional potency of each Ets construct, as measured in the HeLa reconstitution assay, revealed that they all maintained essentially the same level of activity (ϳ22-fold), although BPV-3 was reduced slightly (ϳ15-fold) (Fig. 4B, black bars). With regard to the synergistic response, Pit-1 alone yields a ϳ60-fold response, Ets-1 alone yields a ϳ30-fold response, and together Pit-1 ϩ Ets-1 yield a ϳ525-fold response, which is 5.7 times the sum of the Pit-1 and Ets-1 responses alone (calculated as 90-fold). Similarly, each of these BPV mutant Ets constructs also syn-ergized with Pit-1, yielding 6.2-, 6.7-, and 4 -9-fold synergy for the BPV-1, BPV-2, and BPV-3 constructs, respectively (Fig. 4B,  white bars). Although the synergistic response of BPV-3 is reduced, it is not statistically significant (p ϭ 0.083). Finally, Fig. 4C shows that the expression levels of transiently transfected WT Ets-1 and the BPV mutant Ets-1 proteins were all equivalent. COS-1 cells were utilized to assess Ets protein expression in a background with lower endogenous protein levels. Thus, significant reduction of the physical interaction between Ets-1 and Pit-1 does not appear to interfere with the functional cooperativity of Ets-1 and Pit-1 in vivo, suggesting that DNA contacts and higher order transcription complexes also contribute to these functional responses in the intact cell.
Ets-1/Ets-2 RIII Substitution Abrogates Pit-1 Binding but Not Synergy-To determine better the role of the Ets-1/Pit-1 physical interaction for the synergistic response in HeLa cells, we took advantage of the similarities and dissimilarities of the Ets-1 and Ets-2 proteins which would allow alterations of the amino acids 190 -257 region, yet maintain an intact TAD. Specifically, Ets-1 and Ets-2 are highly related and retain similar protein sequences and functional structures, except in amino acids 180 -300 of the RIII TAD, which are only 6% conserved. Thus, we replaced the Pit-1 interaction face of Ets-1 (amino acids 190 -257) with the corresponding 67-amino acid region of Ets-2, generating a chimeric Ets-1/2 protein that retains an intact RIII TAD but does not physically interact with Pit-1 (3). Indeed, assessment of physical binding of radiolabeled Ets-1, Ets-2, or the Ets-1/2 chimera to the GST-Pit-1 homeodomain (amino acids 199 -291) showed 52, 3, and 6% binding of input, respectively, corroborating our previously published results that binding of Ets-2 to Pit-1 is reduced by 90 -95% (Fig. 5A). Transfection of these various Ets constructs alone revealed that they all activated the rPRL promoter equivalently in the HeLa synergy assay (each ϳ25-fold), verifying that they retained transcriptional activity (Fig. 5B, upper panel, left, black  bars). As shown above, Pit-1 alone resulted in a ϳ85-fold response, WT Ets-1 alone yielded a ϳ30-fold response, and together Pit-1 ϩ Ets-1 yielded a ϳ730-fold response. Thus, the

FIG. 3. Effect of Ets-1 mutations on
Pit-1 homeodomain binding, Ras responsiveness, and synergistic activation of the rPRL promoter. A, structure/function map of chicken p68 c-Ets-1. The Pit-1 interaction face in the RIII TAD is shown as an expanded sequence with the location of the overlapping BPV mutations (DTYRYIESPATKCASNVIPAKE-DPYA, AU1 epitope is underlined) represented as lines. B, in vitro translated p68 Ets-1 or Ets-1 BPV mutations were added to GST and GST-Pit-1 homeodomain. Ets/Pit-1 composites were separated by SDS-PAGE, and the bound 35 S-labeled proteins were quantified by PhosphorImager analysis. Nonspecific binding to GST alone was subtracted from total binding, and specific binding is expressed as percent input.
functional cooperativity of Pit-1 with WT Ets-1 yielded ϳ6.2fold synergistic response, whereas the WT Ets-2 and the Ets-1/2 chimera yielded similar responses of ϳ650-fold, yielding 6and 5.6-fold synergy, respectively (Fig. 5B, upper panel, right). Finally, using an Ets-1/2-specific antibody in Western blot analysis of extracts derived from cells transiently transfected with these various Ets constructs, we found that Ets-1, Ets-1/2, and Ets-2 are each expressed in equivalent amounts (Fig. 5B,  lower panel). Taken together, these data indicate that syner-gistic activation of the rPRL promoter by Ets-1 and Pit-1 is defined primarily by the assembly of their respective TADs at the composite element and that physical interaction of Ets-1 and Pit-1 is not required for this response.
Various TADs Linked to EtsDBD Can Mediate Synergy with Pit-1-Given that physical interaction is not required for the Pit-1/Ets-1 synergy, we next tested the possibility that synergy is determined by the number and/or identity of TADs assembled on the EBS/FPIV composite element. We have reported previously that the Ets-1 RIII TAD is required because an amino-terminal deletion construct that retains this domain (Ets-1⌬5-4) synergizes with Pit-1, but further deletion to remove this domain (Ets-1⌬5-6) results in the loss of both basal activity and synergy with Pit-1 (3). Replacement of Ets-1 RIII with Ets-2 RIII suggests that other Ets factor TADs can replace the Ets-1 RIII TAD. However, this also raises the question of whether any TAD can replace the RIII TAD of Ets-1. To determine whether synergy is governed by an assembly of selective TADs, we linked either the Gal4 or VP16 TADs to various Ets-1 amino-terminal truncations that: 1) retained the minimal TAD/ synergy domain (Ets-1⌬5-4), 2) were devoid of this domain (Ets-1⌬5-6), or 3) retained only the EtsDBD (VP16EtsDBD) (Fig. 6A). Again, as previous studies have shown, Ets-1 strongly synergizes with Pit-1 (6.1-fold synergistic response), the ⌬5-4 construct retained essentially WT synergy levels (5.6-fold synergy), and the ⌬5-6 construct neither activated transcription of PRL alone, nor synergized with Pit-1 (3). Addition of the Gal4 TAD to the ⌬5-4 construct had no effect on either the basal or synergistic responses of the rPRL promoter, despite the additional TAD. In contrast, addition of the Gal4 TAD to the ⌬5-6 construct restored its ability both to activate the PRL promoter (ϳ15-fold, Fig. 6A, black bars) and synergize with Pit-1 (4.3-fold synergy). Although there is a different magnitude in overall activation by the Pit-1/Gal4AD⌬5-6 combination compared with Pit-1/Ets-1, the fold synergies are not significantly different. Transfection of the Ets-1 DBD, which has increased affinity for the Ets binding site because of the deletion of the upstream inhibitory domain, reduced the Pit-1 activation of rPRL promoter activity to 20% of the activity seen in the presence of Pit-1 alone. Thus, it acted as a dominantnegative regulator of the PRL promoter. However, addition of the VP16 TAD to the EtsDBD construct reversed its dominantnegative activity, restoring transcriptional activity to the Ets-DBD and mediating a 5.9-fold synergistic response in combination with Pit-1, similar to the 6.1-fold synergy of the WT Ets-1 construct (Fig. 6A). It is important to note that Ets-1, ⌬5-4, ⌬5-6, and EtsDBD progressively delete the Pit-1 interaction surface, with ⌬5-6 and the EtsDBD lacking this interaction face completely. Thus, these data separately verify that the Pit-1 interaction face is not required for the synergistic response. To verify that these TAD fusion constructs failed to bind Pit-1, GST pull-down studies were performed. Although [ 35 S]methionine labeling of VP16EtsDBD resulted in a poorly labeled protein because of its smaller size and lower methionine content, these data show that binding of VP16EtsDBD and of GalAD⌬5-6 to Pit-1 was significantly reduced at 2 and 3% of input, respectively, compared with Ets-1 (27% of input) (Fig. 6B). Taken together, these data reveal that synergistic activation of the PRL promoter by Ets-1 and Pit-1 can be conferred by the assembly of multiple transcriptional activation domains at the composite element.
A Hierarchy of Cooperating TADs Is Defined by the Pit-1 TAD-Studies of the functional domains of Pit-1 have shown that amino acids 1-80 located at the amino-terminal region of Pit-1 represent its dominant TAD (33). However, previous studies have shown that even with deletion of this putative After 24 h the cells were harvested, and 100 g of total cellular protein from each sample was analyzed by Western blotting utilizing a primary antibody directed against the carboxyl terminus of human Ets-1 (PA-94). The blot was analyzed using pico-West chemiluminescent substrate (Pierce) and autoradiography.
Pit-1 TAD, Pit-1 is able to stimulate synergistic activation of the rPRL promoter with Ets-1 (13). This is in direct contrast to the current studies, which suggest that synergy is conferred by the assembly of multiple TADs at the composite element. Thus, we asked whether the Pit-1⌬2-80 construct, devoid of the Pit-1 TAD, could synergize with the Gal4 TAD and VP16 TAD constructs in the same manner as the full-length Pit-1 construct. We transiently transfected HeLa cells with the Ϫ425PRL promoter and mammalian expression vectors for Pit-1 or Pit⌬2-80 alone and in combination with Ets-1⌬5-6, Gal4AD⌬5-6, Ets-DBD, VP16EtsDBD, VP16Ets280 -440, and VP16Ets330 -440 (Fig. 7A). To address the possibility that the insertion position of the VP16/Gal4 TAD relative to the EtsDBD and the presence of the inhibitory domains flanking the DBD might affect its function (Fig. 3), we chose several available VP16-Ets fusion constructs that placed the VP16 TAD both amino-and carboxyl-terminal to the EtsDBD and contained only the amino-terminal or both inhibitory domains (34). In the absence of Ets-1, Pit-1 stimulated the Ϫ425PRL promoter by 87-fold, and Pit-1⌬2-80 stimulated 41-fold. As described previously, both Pit-1 and Pit-1⌬2-80 cooperated functionally with Ets-1, resulting in 3.6 Ϯ 1-and 4.4 Ϯ 1-fold synergy levels, respectively (Fig.  7A). The synergistic activation of Pit-1 and Pit⌬2-80 with Gal4AD⌬5-6 was approximately equivalent (1.9 Ϯ 0.5-and 3 Ϯ 1-fold synergy, respectively), although the cooperativity of these Pit-1 proteins with Gal4AD⌬5-6 was slightly lower than with full-length Ets-1 (3.6 -4.4-fold synergy). Conversely, the cooperative effects of Pit-1 and Pit⌬2-80 with VP16EtsDBD were notably different, resulting in 7.1 Ϯ 0.9-fold and 2.9 Ϯ 0.8-fold synergistic responses, respectively (Fig. 7A). Fusion of the VP16 TAD to the carboxyl-terminal end of an EtsDBD that retains a portion of the In1 domain (Fig. 3A), resulted in a similar synergistic response (8.6 Ϯ 2.9-fold) compared with VP16EtsDBD (7.1 Ϯ 0.9-fold). Like VP16EtsDBD, the synergy level of VP16Ets280 -440 was greater than that seen with the Gal4AD⌬5-6, indicating that the inclusion of the In1 region is not responsible for the reduced activity of the Gal4AD fusion. Nevertheless, the differential cooperative effects of Pit-1 and Pit⌬2-80 with VP16Ets280 -440 were maintained (8.6 Ϯ 2.9versus 2.7 Ϯ 0.9-fold, respectively. Finally, using a human version of the EtsDBD lacking the In1 inhibitory region with the VP16TAD linked to the carboxyl terminus (VP16Ets330 -440), we recapitulated the strong synergistic response (19 Ϯ 8-fold) and differential Pit-1/Pit⌬2-80 cooperativity (19 Ϯ 8.1versus 3.4 Ϯ 0.8-fold) noted with the chicken Ets VP16EtsDBD (Fig. 7A). As shown in Fig. 7B, the various Ets fusion constructs, particularly those that fail to synergize, are expressed at levels that are equivalent to or greater than Ets-1, and the Pit-1 constructs are also expressed at equivalent levels. These data indicate that the enhanced synergy noted with the VP16 TAD requires the amino-terminal TAD in Pit-1. Although these data support our hypothesis that assembly of multiple TADs at the composite element is required for synergistic activation of the rPRL promoter, they also demonstrate that the identity of the specific TAD combinations modulates the level of the synergistic response.
Distinct Subregions of the Pit-1 TAD (2-80) Mediate Synergy and Ras Responsiveness-To gain further insights into the FIG. 5. Effect of Ets-1/2 RIII substitutions on binding to the Pit-1 homeodomain and synergistic activation of rPRL promoter. A, In vitro translated Ets proteins were added to GST and GST-Pit-1 homeodomain. Ets/Pit-1 composites were separated by SDS-PAGE, and the bound 35 S-labeled Ets-1 was quantified by PhosphorImager analysis. Nonspecific binding to GST alone was subtracted from total binding, and specific binding is expressed as percent input. B, upper panels, HeLa cells were transiently cotransfected with 75 ng of promoter and 1 ng of pRLC-Renilla. 50 ng of pSG5-Ets constructs and 50 ng of pRSV-Pit-1 were added where indicated. After 24 h, the cells were assayed for luciferase activity. Bars represent the mean Ϯ S.E. of triplicate samples in three transfections. Numbers indicate fold synergy Ϯ S.D. Lower panel, COS-1 cells were transiently transfected with 10 g of the various Ets-1 constructs. After 24 h the cells were harvested, and 100 g of total cellular protein from each sample was analyzed by Western blot analysis utilizing a primary antibody directed against amino acids 362-374 of human Ets-1, which detects Ets-1 and 2 (Santa Cruz Biotechnology). The blot was analyzed using ECL Plus (Amersham Biosciences) and the Storm system (Amersham Biosciences). structure/function mechanisms of the Pit-1 TAD responsible for the Ras-and the Pit-1/Ets-1 synergistic response of the rPRL promoter, we tested WT and two amino-terminal deletions encompassing the Pit-1 TAD (Fig. 8, A and B, WT, ⌬2-80, ⌬2-45) for their ability to enhance the Ras response (Fig. 8C) and to mediate Pit-1/Ets-1 synergy (Fig. 8D). The data reveal that WT Pit-1 enhances the 10-fold Ras response to 35-fold, whereas the ⌬2-80 construct fails to enhance the Ras response and yet the ⌬2-45 construct retains WT levels of enhancement of the Ras response to ϳ35-fold (Fig. 8C). These results effectively map the Ras-responsive subdomain of the Pit-1 TAD to amino acids 46 -80. Of note, the lack of enhancement of the ⌬2-80 construct is not the result of lack of protein expression because these various Pit-1 constructs were expressed in a similar manner (Figs. 7B and 8B). In comparison, the Pit-1 ⌬2-80 construct was equally as efficient at promoting synergy with Ets-1 as WT Pit-1 (Fig. 8D), indicating that the entire TAD is dispensable in the synergistic response (as shown in Fig. 7). Curiously, although the ⌬2-45 construct exhibits higher basal activity (ϳ600-fold) compared with WT (250-fold), its ability to synergize with Ets-1 is compromised and evinces only a 1.8-fold synergy compared with 4.4-and 3.8-fold synergistic responses for the WT and ⌬2-80 Pit-1 constructs, respectively. Finally, Ets-1 expression is not altering the levels of WT and amino-terminally truncated Pit-1 proteins expressed (Fig. 8B).

Interplay of Cis-and Trans-elements in the Regulation of Synergy and Ras
Signaling-Pit-1 has been identified as a pituitary-specific signal integrator that directs both positive and negative regulatory stimuli to genes in pituitary somatotropes, lactotropes, and thyrotropes. This signal integration appears to be regulated by the binding of Pit-1 to a variety of composite cis-elements as well as functional interactions with other nuclear regulatory factors. In the current study, we examined the spacing and organization of the individual sites within the EBS/FPIV composite element, the requirement of physical interaction between Pit-1 and Ets-1, and the role of distinct TADs and their physical arrangement which allow for targeting of the rPRL promoter by the Ras signaling pathway and which promote the synergistic response to Ets-1 and Pit-1.
Here, we show that although the Pit-1/Ets-1 composite element tolerates a high degree of flexibility and that direct physical interaction between Pit-1 and Ets-1 is not required for transcription synergy, an assembly of specific TADs on the composite element is required for optimal synergistic activation of the rPRL promoter. Taken together, these data not only provide important information regarding transcription synergy in general, but also reveal novel mechanistic insights regarding the precise cis-and trans-requirements for transcription synergy for the rPRL promoter. Specifically, the data show a differen- Bars represent the mean Ϯ S.E. of triplicate samples in three transfections. Numbers indicate fold synergy. B, in vitro translated p68 Ets-1 or Ets TAD substitutions were added to GST and GST-Pit-1 homeodomain. Ets/Pit-1 composites were separated by SDS-PAGE, and the bound 35 S-labeled proteins were quantified by PhosphorImager analysis. Nonspecific binding to GST alone was subtracted from total binding, and specific binding is expressed as percent input. tial use of distinct TAD subtypes (e.g. Ets-1, Pit-1, Gal4, VP16) and TAD subregions (e.g. Pit-1 TAD R1 versus R2) to mediate either synergy or Ras responsiveness. Finally, because the Pit-1 R2 TAD subregion lacks a proline-directed MAP kinase site, it appears to be a novel Ras-responsive TAD.
Pit-1 Monomer versus Dimer DNA Binding Sites Are Not a Primary Determinant for Ets-1 Synergy or Ras Responsiveness-POU transcription factors are allosterically regulated by the sequence and organization of their DNA binding sites; therefore, the specific DNA-induced conformation of Pit-1 appears to select functional interactions with specific coregulatory proteins generating transcription complexes that mediate distinct responses (25,27,35). In the studies shown here, both the monomeric 1D site and the palindromic FPI dimer site serve nearly as well as the endogenous FPIV Pit-1 binding site in mediating the Ras response in GH4 pituitary cells, whereas the 1D monomer site is preferred for transcriptional synergy in HeLa cells (Fig. 1). These data are consistent with a previously published report showing that intact Pit-1 displayed optimal synergy with the ER, regardless of the identity of the Pit-1 site at a composite Pit-1/ER element from the rPRL promoter (26). Additionally, the palindromic FPI dimer site used here has been reported to present Pit-1 in a conformation that preferentially binds the CBP coactivator (27,35). Yet our data reveal that for the Ras response, the precise DNA structure of the Pit-1 binding site as a monomer or dimer, and therefore the precise conformation of Pit-1 appears to be quite resilient. In contrast, the ϳ30% enhancement of the Pit-1 plus Ets-1 synergistic response of the one-dimensional monomer reporter construct is likely the result of the improved Ets-1 stimulation (Fig. 1C). The WT behavior of the EBS and FPIV site-specific mutants in the synergistic response suggests that additional Ets-1 and Pit-1 sites in the Ϫ425 rPRL promoter function in a redundant manner in this assay, particularly those found in FPIII and the EBS at position Ϫ76 adjacent to FPI at Ϫ65 (see Fig. 1A, top panel). Finally, it is important to note that these data do not preclude a preferred Pit-1 conformation for the responses tested, but rather the data indicate that of the var- 2 g of pSG5-EtsDBD, 3.5 g of pSG5-VP16 EtsDBD, 10 g of pSG5-Ets-1, 3.3 g of pSG5 Ets-1⌬5-6, 4.7 g of pSG5-Gal4AD⌬5-6, 10 g of pSG5-Ets-1⌬5-4, and 10 g of pSG5Gal4AD⌬5-4. After 24 h the cells were harvested, and 100 g of total cellular protein from each sample was analyzed by Western blot utilizing a primary antibody directed against the carboxyl terminus of human Ets-1 (PA-94). The blots were analyzed using pico-West chemiluminescent substrate (Pierce) and autoradiography or ECL Plus and the Storm system, respectively. Because this antibody also detects endogenous Ets factors the specific expressed protein in each lane is indicated with an asterisk. Right panel, HeLa cells were transiently transfected with 50 ng of pCGN2-Pit-1 and 12.5 ng of pCGN2-Pit-1⌬2-80. After 24 h the cells were harvested, and 100 g of total cellular protein from each sample was analyzed by Western blot utilizing a primary antibody directed against the hemagglutinin tag (Santa Cruz Biotechnology). The blot was analyzed using pico-West chemiluminescent substrate and autoradiography.
ious Pit-1 binding sites analyzed thus far, all are functional in the Ras response and synergy assays.
The Spatial Requirements for the Ets-1/Pit-1 Composite Element Are Flexible-The spatial flexibility of an EBS relative to the binding site of its vicinal partner has been examined in detail in only a few composite cis-acting elements (19,36,37). The best studied is the serum response element, in which randomized DNA sequences were inserted upstream of a canonical SRF binding sequence (CArG box), and after several rounds of electrophoretic mobility shift assay selection for ternary complex formation, it was found that EBS sequences were enriched in the upstream region. In this in vitro gel shift assay, both the spacing and orientation of the EBS sequences relative to the SRF site in the serum response element could be altered substantially without affecting the efficiency of ternary complex formation (19). By contrast, using functional reporter gene assays, certain artificially constructed Ets composite elements (e.g. Ets/AP-1 or Ets/Ets) displayed a very limited flexibility with respect to spacing and/or orientation, when tested for basal activity and Ras responsiveness (36,37). The natural separation between the EBS and FPIV is about 8 bp ( Fig. 2A), and here we found that both the Ras and synergistic responses of the rPRL promoter were reduced significantly only when the spacing was increased from 28 to 32 bp, suggesting a relatively flexible configuration of the EBS and FPIV sites. Additionally, functional responses did not appear to correlate with the helical phasing of the binding sites relative to each other (Fig. 2, B  and C), and reversing the orientation of the Pit-1 binding site also failed to decrease functional responses (data not shown). These spacing data, together with those in the literature (19,36,37), raise the question as to which parameters actually define a composite element. For example, the spacing limit of 28 and 32 bp for EBS and FPIV found here is in agreement with the 23-26-bp limit found for the EBS and serum response element in the in vitro ternary complex formation studies (19). However, these results are in contrast to other functional data, showing that spacing and orientation of an EBS relative to either an AP-1 or other EBS sites are quite limited (36,37). Further challenging the definition of an EBS/Pit-1 composite element is the observation that the ancestrally related rat growth hormone gene promoter also contains Pit-1 binding sites as well as putative Ets binding sites within 32 bp, yet fails to display either Ras or synergistic responses (10,11). Taken together, these data indicate that promoter context plays a significant role in defining a composite element. Thus, the overall DNA sequence composition of the promoter (AT-versus GC-rich), the position of the composite element relative to other cis-sites, the sequence identity of each component of the composite element, and the specific proximal promoter structure (e.g. CAAT, TATA, and initiator sequences), may all contribute to the maximal synergistic activity of any putative composite element (27, 38 -40).
Physical Binding of Ets-1 to Pit-1 Is Not a Primary Determinant for Ras Responsiveness or Synergy-The assembly of transcription factors on composite elements is typically driven by vicinal DNA sites that bind these factors via protein-DNA interactions and stabilized via protein-protein interactions (37,41,42). We have shown previously by GST interaction assays and NMR analysis that the homeodomain of Pit-1 (amino acids 199 -291) interacts directly with the RIII TAD of chicken c-Ets-1 (amino acids 190 -257), albeit weakly (K d ϳ300 M) (12, 13). Here we show by three separate approaches for altering FIG. 8. Role of subdomains of Pit-1TAD in Ras and Ets-1/Pit-1 synergy responses. A, illustration of the functional domains of Pit-1 and amino-terminal deletion constructs. B, HeLa cells were transiently cotransfected with 3 g of promoter and 100 ng of prh-TK-Renilla. 5 g of pSG5-Ets-1 and 50 ng of pCGN2-Pit-1 or 12.5 ng of pCGN2-Pit-1⌬2-80, or 100 ng of pCGN2-Pit-1⌬2-45 were added where indicated. DNA was held constant with the addition of pCMV-␤-galactosidase. After 24 h, the cells were harvested, and 100 g of total cellular protein from samples was analyzed by Western blot for expression of the Pit-1 constructs utilizing a primary antibody directed against the hemagglutinin tag. The blots were developed using pico-West chemiluminescent substrate and autoradiography. C, GH4T2 cells were transiently cotransfected with 3 g of promoter and 100 ng of prh-TK-Renilla. 2 g of pSV-Ras and 50 ng of pCGN2-Pit-1 or 12.5 ng of pCGN2-Pit-1⌬2-80, or 100 ng of pCGN2-Pit-1⌬2-45 were added where indicated. DNA was held constant with the addition of appropriate control vectors. After 24 h, the cells were harvested and assayed for luciferase activity. Bars represent the mean Ϯ S.E. of three samples in three transfections utilizing at least two different plasmid preparations. The dotted line indicates level of Ras response in the absence of transfected Pit-1. D, HeLa cells were transfected as described in B. After 24 h, the cells were harvested and assayed for luciferase activity. Bars represent the mean Ϯ S.E. of three samples in three transfections utilizing at least two different plasmid preparations. Numbers represent the mean fold synergy. Asterisks signify a value that differs significantly from that elicited by the corresponding WT construct (p Ͻ 0.05 by Student's t test).
the Pit-1 interaction motif in Ets-1, a BPV scanning mutagenesis scheme, an Ets-2 substitution method, and a deletion approach, that despite significant loss of physical interaction (ϳ 90%) in vitro, these mutants retained essentially WT levels of Ras responsiveness and transcriptional synergy in vivo (Figs. 3-6). One possible interpretation of these data is that the in vivo reporter assays are measuring a complex array of interactions that include chromatin reorganization, complexity of the promoter, transcription factor-DNA and transcription factor-coactivator interactions, making it difficult to identify the functional impact of a single type of interaction, whereas the GST pull-down assay is limited to transcription factor-transcription factor interactions. Alternatively, another interpretation is that direct physical interaction of Ets-1 and Pit-1 is not critical for these responses in vivo.
Optimal Synergy Is Conferred by the Assembly of Select TADs-In general, the key mechanism of action of composite elements appears to be the localized assembly of multiple TADs, generating several juxtaposed activation surfaces that recognize transcriptional coactivators better (43)(44)(45). However, the precise contribution and specificity of each TAD for the synergistic response previously have not been examined systematically. Here we show that in the transcriptional synergy assay, retention of a single TAD linked to Ets-1 is sufficient to reconstitute Ets-1/Pit-1 synergy, i.e. the Pit-1 amino-terminal TAD is not required for synergistic responses (Figs. 7 and 8). This suggests that a secondary TAD within the Pit-1 homeodomain may be responsible for the synergistic response.
Moreover, we also demonstrate that the three different classes of TADs represented by Ets-1, Gal4, and VP16 generate distinct hierarchies of cooperativity with Pit-1, based on whether intact Pit-1 or Pit-1⌬2-80, devoid of its TAD, is used as the Ets-1 partner. Specifically, for intact Pit-1, the TAD rank order for synergistic activation is: VP16 Х Ets Ͼ Gal4. In contrast, for Pit-1⌬2-80, the TAD rank order is: Ets Ͼ Gal4 Ͼ VP16 (Figs. 6 and 7). These data show that the Pit-1 TAD (amino acids 2-80) is required for the VP16 TAD to display maximal synergy, yet this same domain of Pit-1 is dispensable for optimal synergy with Ets-1 and Gal4. This differential requirement of the Pit-1 TAD for maximal cooperativity with the VP16 TAD implies that a distinct mechanism of transcription activation is used by the VP16-Ets/Pit-1 combination. One possible explanation for the optimal synergy of the VP16-Ets fusion with full-length Pit-1 (7.1-fold synergy) compared with the less than optimal synergy with Pit-1⌬2-80 (2.9-fold synergy) may be that the VP16 TAD linked to Ets-1 is only able to cooperate with the amino-terminal TAD and is unable to coop-erate with the homeodomain TAD, whereas the Ets-1 and Gal4 TADs appear to be able to cooperate with the homeodomain TAD alone.
The Pit-1 TAD (Amino Acids 2-80) Can Be Subdivided into Modulatory and Ras-responsive Subdomains-The precise role of the Pit-1 amino-terminal TAD (amino acids 1-80) appears to be quite complex. Based on the summary of our data presented in Fig. 9, we propose that amino acids 1-45 function as a modulatory region 1 (R1) and amino acids 46 -80 function as an effector region 2 (R2) (Fig. 9). Specifically, deletion of both R1 and R2 (Pit-1⌬2-80) results in modest basal activity, no Ras response, and essentially a WT synergistic response (Figs. 7-9). Addition of the R2 region alone, however, doubles the basal response, confers the Ras response, but diminishes the synergistic response. Finally, further addition of the R1 region, generating the WT Pit-1, results in a reduction in basal activity, a persistence of the Ras response, and a gain in the strength of the synergistic response. This summary reveals that the R1 region has both negative and positive effects, on the basal and synergistic responses, respectively, whereas the R2 region is required for the Ras response. Functional analysis of the Pit-1 TAD conducted by other investigators has revealed a similar subdivision of function. Specifically, mutation of proline 24 to leucine in the Pit-1 R1 TAD has been shown to inhibit the ability of Pit-1 to interact with CBP (46). Deletional analysis of the amino-terminal Pit-1 TAD has identified a tyrosine-rich region located between amino acids 45 and 72 (R2), similar to that of hLEF-1, which is required for synergistic activation with the ER (26). Phosphorylation of these tyrosines does not appear to be required for synergy with ER because mutation of these tyrosine residues to phenylalanine residues, which are not substrates for phosphorylation, resulted in a mutant Pit-1 that retained full synergistic activation. Although the mechanism for Ras stimulation of the Pit-1 R2 TAD has yet to be elucidated, we know that in Ets-1 the Thr-82 MAP kinase phosphorylation site is required for Ras responsiveness because mutation of this site to an alanine blocks the ability of Ets-1 to enhance the Ras response (10,14). There are potential phosphorylation sites within the Pit-1 TAD, but none of them is a proline-directed MAP kinase site, and to date only serine 115 and threonines 219 and 220 have been shown to be phosphorylated by various signaling pathways and kinases (47,48), suggesting that Ras activation of the amino-terminal Pit-1 TAD is not mediated through direct phosphorylation of the TAD. Thus, the R2 subregion represents a novel and tissuespecific Ras-regulated TAD.
Taken together, the data show remarkable flexibility in the FIG. 9. Summary of basal activity and Ets-1 synergy in HeLa cells and enhancement of Ras responsiveness of GH4T2 cells. Deletion of both R1 and R2 (Pit-1⌬2-80) results in modest basal activity, no Ras response, and essentially a WT synergistic response. Addition of the R2 region alone doubles the basal response, confers the Ras response, but diminishes the synergistic response. Finally, further addition of the R1 region, generating the WT Pit-1, results in a reduction in basal activity, a persistence of the Ras response, and a gain in the strength of the synergistic response. Thus, the R1 region has both negative and positive effects on the basal and synergistic responses, respectively, whereas the R2 region is required for the Ras response. organization of the Ets-1/Pit-1 composite element with regard to both Ras responsiveness and Ets-1/Pit-1 transcription synergy. However, the data also reveal that the Ras and synergistic responses are mediated by the specific assembly of select activation domains on this composite element, indicating that distinct mechanisms underlie these two transcriptional effects.