Peptidylprolyl Isomerase Pin1 Directly Enhances the DNA Binding Functions of Estrogen Receptor α*

Background: Estrogen receptor α (ERα) activity in breast cancer cells is increased by Pin1, an isomerase that alters protein conformation. Results: Pin1-mediated isomerization increases ERα DNA binding affinity. Conclusion: Pin1 allosterically regulates ERα DNA binding directly via isomerization of phosphorylated receptor. Significance: Pin1 regulation of ERα provides a framework for understanding regulation by intrinsically disordered domains on transcription factor function. The transcriptional activity of estrogen receptor α (ERα), the key driver of breast cancer proliferation, is enhanced by multiple cellular interactions, including phosphorylation-dependent interaction with Pin1, a proline isomerase, which mediates cis-trans isomerization of the N-terminal Ser(P)118-Pro119 in the intrinsically disordered AF1 (activation function 1) domain of ERα. Because both ERα and Pin1 have multiple cellular partners, it is unclear how Pin1 assists in the regulation of ERα transactivation mechanisms and whether the functional effects of Pin1 on ERα signaling are direct or indirect. Here, we tested the specific action of Pin1 on an essential step in ERα transactivation, binding to specific DNA sites. DNA binding analysis demonstrates that stable overexpression of Pin1 increases endogenous ERα DNA binding activity when activated by estrogen but not by tamoxifen or EGF. Increased DNA binding affinity is a direct effect of Pin1 on ERα because it is observed in solution-based assays with purified components. Further, our data indicate that isomerization is required for Pin1-modulation of ERα-DNA interactions. In an unbiased in vitro DNA binding microarray with hundreds of thousands of permutations of ERα-binding elements, Pin1 selectively enhances the binding affinity of ERα to consensus DNA elements. These studies reveal that Pin1 isomerization of phosphorylated ERα can directly regulate the function of the adjacent DNA binding domain, and this interaction is further modulated by ligand binding in the ligand-binding domain, providing evidence for Pin1-dependent allosteric regulation of ERα function.

porates kinase and protein-protein interactions via the AF1 domain, ligand binding and coregulator interactions at the LBD, and DNA interactions at the DBD.
The AF1 domain of ER␣ is an intrinsically disordered (ID) domain. Although ID domains lack a specific identifying motif or secondary structure, they play key roles in post-translational modifications, protein-protein interactions, and allosteric control (19 -21). By virtue of their unstructured characteristics, ID domains are difficult to study, yet AF1 has been shown to provide an interaction surface for coregulator proteins (10,(22)(23)(24)(25). An understanding of the interactions in ID domains could provide a mechanistic window to better understand the allosteric role of unstructured domains in transcriptional signaling (26).
Our studies indicate that high levels of Pin1 expression are associated with worse outcomes in ER␣ ϩ breast tumors as well as increased ER␣ signaling and transcriptional activity (25,31). Pin1 is recruited to phosphorylated ER␣ at Ser 118 in response to estrogen, tamoxifen, and EGF (25). Pin1 enhances ER␣-mediated transcription induced by estrogen and tamoxifen and increases ER␣ dimerization. These previous studies also showed that the Pin1-mediated increase in ER␣ transcriptional activity was mediated through the AF1 domain (25). Collectively, these studies establish Pin1 as an AF1 binding partner that enhances ER␣ transcriptional function.
Multiple potential mechanisms have been proposed to describe Pin1 regulation of ER␣ transcriptional function, highlighting the complexity of Pin1 activity on ER␣ signaling. Like ER␣, Pin1 binds multiple components of transcriptional complexes, including RNA polymerase II, histones, general transcription factors, and ER␣ coactivators and corepressors (34 -38). Moreover, Pin1 alters signaling in breast cancer cells that can indirectly influence ER␣ activity (34, 36, 39 -41). Hence, traditional cell-based approaches to elucidate Pin1 activity on ER␣-mediated transcription are confounded by the potential for both direct and indirect effects. Therefore, to focus specifically on direct effects of Pin1 on ER␣ function, we examined how Pin1 binding and isomerization might affect ER␣ DNA binding activity. To enable direct analysis of DNA binding affinity and sequence specificity, we used purified proteins and an unbiased approach on the scale of 385,000 DNA sequences, providing a mechanistic readout of Pin1-mediated effects on ER␣ DNA binding in the process. Together, the data provide novel insights into the roles of phosphorylation and Pin1 on ER␣ transcriptional regulation and describe a new approach for refined definition of components within large transcription factor complexes.

Experimental Procedures
Cell Culture and Treatments-MCF7 cells and MCF7 cells stably overexpressing GFP-Pin1 and GFP (25) were maintained in DMEM (Life Technologies, Inc.), 10% fetal bovine serum (FBS; Hyclone), and 1% penicillin-streptomycin (Life Technologies) at 37°C and 10% CO 2 . In experiments involving treatments with 0.1% EtOH (vehicle), 10 nM 17␤-estradiol (E 2 ; Steraloids, Inc.), or 100 nM 4-hydrotamoxifen (OHT), cells were placed in estrogen-depleted phenol-red free medium consisting of 10% dextran-charcoal-stripped FBS for 3 days prior to the addition of hormone or vehicle. For EGF treatments, the medium was changed to phenol red-free DMEM or Opti-MEM (Life Technologies) overnight prior to treatment. The treatments were carried out for the times indicated in the figure legends.
Western Blot and Antibodies-Western blot was performed as described (25). ER␣ antibody (Santa Cruz Biotechnology, Inc.) was used to detect purified and endogenous ER␣. Antibody directed against Ser(P) 118 ER␣ (Cell Signaling) was used to detect Ser(P) 118 -ER␣. Pin1 antibody (Epitome) was used to detect purified and endogenous Pin1. Detection of actin with an anti-actin antibody (Sigma) was used as loading control.
Electrophoretic Mobility Shift Assay (EMSA)-ERE and mutated ERE (mERE) probes listed in Table 1 were annealed as 32-mer sense (0.5 g) and 13-mer antisense (0.2 g) in solution containing 10ϫ Klenow buffer in 100 l of H 2 O at 90°C and slow cooled to room temperature. An extension reaction was then performed to fill the overhang in a 200-l reaction containing 100 l of annealed oligonucleotide, 4 l of 10 mM dNTP (2.5 mM concentration of each: dATP, dCTP, dGTP, and dTTP), 10 l of 10ϫ Klenow buffer, 5 l of Klenow polymerase (New England Biolabs), and H 2 O. This mixture was incubated at 16°C overnight. Unincorporated dNTPs were removed by passing the mixture through a G-50 column twice. To end-label probes with radioactive 32 P, 20 ng of reannealed oligonucleotides were mixed with 2 l of 10ϫ T3 PNK buffer, 1 l of 0.1 M DTT, 1 l of T4 PNK (New England Biolabs), and the final volume was brought up to 18 l by adding H 2 O. Two l of [␥-32 P]ATP (6,000 Ci/mmol) was then added to the reaction and incubated for 4 h at 37°C. Unincorporated 32 P was removed by passing the reaction through a G-50 column twice. Afterward, TE buffer (10 mM Tris-Cl, pH 8.0, 1 mM EDTA) was added to the purified reaction to a final volume of 50 -100 l and stored at Ϫ20°C. University of Wisconsin (Madison, WI) radioactive safety guidelines were strictly followed during these procedures. Cells were treated as indicated in the figure legends and harvested. Cells were pelleted and resuspended in 100 l of buffer A (10 mM HEPES, pH 7.8, 10 mM KCl, 1.5 mM MgCl 2 ) supplemented freshly with 0.5 mM PMSF, 10 g/ml aprotinin, 0.1% Nonidet P-40, and 1 mM DTT. Cells were then incubated on ice for 30 min and spun down for 1 min at 13,000 rpm and 4°C. The supernatant (cytoplasmic fraction) was moved to a new tube. For nuclei lysis, the pellet was dissolved in 50 l of buffer C (20 mM HEPES, pH 7.8, 450 mM NaCl, 0.2 mM EDTA, 1.5 mM MgCl 2 , 25% glycerol) supplemented with 0.5 mM PMSF, 10 g/ml aprotinin, and 0.5 mM DTT. The lysis reaction was incubated on ice for 30 min and spun down for 10 min at 13,000 rpm and 4°C. Supernatants were transferred to new tubes, and protein concentrations were measured by Bradford assay (Bio-Rad).
The EMSA was performed using fresh nuclear extracts or purified proteins. For cell studies, 10 g of nuclear extract was incubated with 2 l of 5ϫ binding buffer (375 mM NaCl, 75 mM Tris-Cl, pH 7.5, 7.5 mM EDTA, 20% glycerol, 100 g/ml BSA), 1.5 mM DTT, 1 l of poly(dI-dC) (Sigma), and H 2 O was added up to a volume of 9 l on ice. ER␣ supershifts were performed with 2-4 l of ER␣ antibody Ab712 (42). For in vitro EMSA, purified ER␣ (Life Technologies or Pierce) was used in the presence and absence of Pin1, Pin1 K63A, or juglone (Sigma) and incubated at 30°C for 30 min. One l of 32 P-end-labeled probe was added to the reaction and further incubated at room temperature for 20 min. Afterward, samples were loaded onto a native 4% acrylamide gel composed of 8 ml of 30% acrylamide, 2% bis, 1.5 ml of TBE, 600 l of 10% APS, and 60 l of TEMED and electrophoresed at 200 V for 1.5 h in 0.25ϫ TBE. The gel was then placed on Whatman paper, dried in a gel dryer (Bio-Rad) for 45 min, and exposed to x-ray film (Eastman Kodak Co.) at Ϫ80°C for 1-7 days, depending upon the signal. Bands were quantified by exposing the dried gel to a phosphoscreen (Eastman Kodak Co.) and quantified by a PhosphorImager (GE Healthcare).
Specificity and Affinity for Protein (SNAP) Array-A customdesigned DNA array known as the SNAP array (Proteovista LLC) was synthesized using maskless array technology (Roche Nimblegen). ERE variations were tested within the sequence, 5Ј-GCGGTAGGTCA(N 9 )GCTGCGGAGCAGC(N 9 )ЈTGAC-CTACCGC-3Ј, where N is any nucleotide, NЈ is the reverse complement, and the 15-mer core sequence is in boldface type. Control probes included response elements for androgen receptor and glucocorticoid receptor as well as mutated EREs. DNA hairpins were induced on an array, as described previously (43). A hybridization chamber was applied over the array, and the array was hydrated twice with deionized water. The array was then blocked with 2.5% milk for 1 h at room temperature on a rotator (2-3 rpm). After the blocking solution was removed from the array, the transcription factor binding buffer (Proteovista LLC) was used to rinse the array. Purified ER␣ (Life Technologies), preincubated with or without a 10-fold molar excess of Pin1, was applied to the array in buffer plus 1 M E 2 , incubated for 1 h with rotation, and detected with an Alexa Fluor 647-labeled anti-ER␣ antibody (Santa Cruz Biotechnology). After incubation, the array was rinsed briefly, dried, and scanned using a GenePix 4000B scanner (Molecular Devices).
Arrays were performed in duplicate. The data from the arrays were extracted using NimbleScan 2.4 (Roche Applied Science), and the data were processed as described previously (44,45).
Fluorescence Polarization (FP)-SNAP microarray results were verified using individual DNA sequences in FP assays. Dilutions of ER␣, either alone or with a 10-fold molar excess of Pin1 or Pin1 starting concentration, as noted in the legend, were preincubated at 30°C for 20 min in a 96-well plate (Corning 3676). Fluorescein-labeled, hairpinned DNA (IDT Technologies; core sequences shown in Fig. 5) was added to a final concentration of 1 nM, and the plate was incubated for an additional 10 min at 30°C. To avoid quenching of the fluorescein label by the 5ЈG of the array DNA probe, a T nucleotide was included at the 5Ј-end of the 53-mer DNA sequences (Seq1, Seq2, and Seq3). The mERE had the sequence, 5Ј-TGCGGTG-AGACACAGGGTTTCGCTGCGGAGCAGCGAAACCCT-GTGTCTCACCGC-3Ј, where the mutated ERE is in boldface type. The core sequence that was altered between different probes is underlined within the adjacent backbone separated by a GGA loop in the hairpin. FP was measured using a Safire2 plate reader (Tecan), and the data were analyzed by sigmoidal dose response (variable slope) using Prism (GraphPad) to yield the dissociation constant (K d ).
Chromatin Immunoprecipitation (ChIP)-MCF7 cells overexpressing Pin1 or vector were treated with E 2 for the indicated times periods, followed by chromatin cross-linking, lysis, fragmentation, immunoprecipitation, and quantitative real-time RT-PCR as described previously (33,46). ER␣ was immunoprecipitated using ER␣ (HC-20, SC-543) antibody (Santa Cruz Biotechnology). Quantitative real-time RT-PCR was carried out with 1 l of input or 4 l of immunoprecipitated sample and 200 nM forward and reverse primers (primer sequences available upon request). Data are calculated as percentage of input by the ⌬⌬Ct method.
Protein Purification-GST-tagged Pin1 and K63A Pin1-expressing plasmids (gifts from Dr. Kun Ping Lu, Harvard Medical School) were transformed into Escherichia coli BL21 (pLysS) cells (Novagen). Proteins were purified as described previously (29). To cleave the GST tag, proteins were incubated with thrombin (Sigma) solution overnight at 4°C. The eluted fractions were combined and buffer-exchanged with 50 mM Tris-Cl, pH 8.0, 200 mM NaCl, and 10% glycerol using 10,000 kDa cut-off centrifugal filters (Millipore). The level of protein purity was assessed using SDS-PAGE stained with Coomassie Brilliant Blue (Bio-Rad).
In Vitro Dephosphorylation Assay-This assay was carried out as described previously with modifications (25). Purified ER␣ (Life Technologies) was incubated with 50 ng of purified trans-specific PP2A holoenzyme (a gift from Dr. Yongna Xing, University of Wisconsin (Madison, WI)) and 1 g of purified BSA (control) or Pin1 in dephosphorylation buffer (50 mM Tris-Cl, pH 7.5, 50 mM NaCl, 50 M MnCl 2 ) at 30°C for 0 -90 min. The reaction was terminated with 2ϫ SDS sample buffer at the indicated times, and phosphorylation was assessed by Western blot using Ser(P) 118 ER␣-specific antibody. Equivalent loading was confirmed by reprobing with ER␣-specific antibody.
Statistical Analysis-Student's paired t tests and analysis of variance were performed using GraphPad Prism (GraphPad Software). p values of less than 0.05 were considered significant and are indicated by asterisks in the figures.

Results
Pin1 Enhances the DNA Binding Activity of ER␣ from MCF7 Breast Cancer Cells in the Presence of Estrogen-Our previous studies indicated that Pin1 catalyzed conformational shifts in the ER␣ N-terminal AF1 domain, leading to enhanced transcriptional activity (25). Phosphorylation can also functionally alter the DBD and LBD of ER␣ (16,47). Thus, we hypothesized that Pin1-induced changes in AF1 may alter receptor transcrip-tional activity through other functional domains. We chose to focus on the DBD, which is immediately adjacent to the AF1 domain. ER␣ is maximally phosphorylated within 30 min by E 2 treatment in our MCF7 model system (Fig. 1A). To evaluate the role of Pin1 in endogenous ER␣ DNA binding function, MCF7 breast cancer cells stably expressing GFP-Pin1 or GFP alone (vector) were stimulated with E 2 for 30 min. EMSA with nuclear extracts showed that Pin1 overexpression increased ER␣ DNA binding activity in the presence of ligand by ϳ5-fold (Fig. 1, B and C). The Pin1-induced increase in DNA binding affinity concurs with increased transcriptional activity of liganded ER␣

. Pin1 overexpression increases ER␣-ERE interaction in breast cancer cells.
A, purified recombinant ER␣ (rER␣; 0.075-0.3 pmol, 5-20 nM) and MCF-7 cell extracts treated with EtOH or 10 nM E 2 (0 -480 min) were separated by SDS-PAGE, and Western blot was performed for Ser(P) 118 -ER␣, ER␣, and actin using appropriate antibodies. B, nuclear extracts from MCF7 cells stably overexpressing GFP (vector (V)) or GFP-Pin1 (Pin1) treated without (Ϫ) and with 10 nM E 2 (E2) were incubated with 32 P-labeled ERE followed by gel electrophoresis and autoradiography. SS, ER␣ supershift using anti-ER␣ antibody. ERE represents unbound free probe. C, EMSA autoradiography was quantified using a PhosphorImager, and data are represented as means Ϯ S.E. (error bars) for at least three independent experiments. *, p Ͻ 0.05 comparing E 2 -treated vector and Pin1-expressing cells. Data were normalized to vehicle-treated vector-expressing cells. D, Western blot analysis was performed for ER␣ and Pin1 on 10-g nuclear extracts of E 2 -treated and untreated MCF7 cells stably overexpressing GFP (vector (V)) or GFP-Pin1 (Pin1) using appropriate antibodies. E, EMSA was performed as in B in vector-or Pin1-expressing cells treated with (ϩ) or without (Ϫ) 100 nM OHT. F, EMSA for E was quantified as in C, and data are represented as means Ϯ S.E. for at least three independent experiments. n.s., not significant comparing OHT-treated vector and Pin1-expressing cells. Data were normalized to vehicle-treated vector-expressing cells.
induced by Pin1 overexpression in MCF7 (25). Because Pin1 can regulate ER␣ protein abundance levels (31), Western blot analysis was performed to confirm that ER␣ protein levels were equivalent in lysates used in EMSAs (Fig. 1D). These data indicate that Pin1 can increase DNA binding independent of its activity on receptor protein levels. The increase in DNA binding in E 2 -treated cells implicates a role for ligand and the LBD. Hence, similar experiments were conducted in MCF7 cells treated with OHT ( Fig. 1, E and F). In contrast to E 2 , Pin1 did not significantly enhance the ER␣-DNA interaction in the presence of OHT. Together, these data indicate that Pin1 regulates ER␣ DNA binding in a ligand-specific manner, independent of effects on ER␣ protein levels.
EGF activates ER␣ by kinase-mediated phosphorylation of the AF1 domain and promotes Pin1 binding to the receptor (12,25,48). To further explore the specificity of Pin1 regulation on ER␣ DNA binding, EMSA was conducted with EGF-treated cells. EGF-induced phosphorylation occurs on a different time scale than E 2 , and maximal Ser 118 phosphorylation occurred within 5 min of stimulation ( Fig. 2A). Thus, MCF7-GFP and MCF7-Pin1 GFP cells were treated with EGF for 5 min, and nuclear extracts were prepared. EMSAs showed that Pin1 increased DNA binding activity in the absence of stimulation (Fig. 2, B and C). However, EGF treatment resulted in no further enhancement of the Pin1 effect on ER␣ DNA binding function. Because previous studies showed that EGF treatment was sufficient to promote Pin1 binding to ER␣ (25), these data suggest that EGF-induced phosphorylation and recruitment of Pin1 to ER␣ is not sufficient for Pin1-mediated effects on ER␣ DNA binding activity.
Pin1 cis/trans-Isomerase Activity Is Required to Increase ER␣ DNA Binding Function-To better define Pin1 activity on ER␣ DNA binding, solution-based DNA binding assays using purified components were conducted. We used commercially purified ER␣ that is phosphorylated at Ser 118 (Fig. 1A). Recombinant ER␣ binds to a perfect ERE but not to the mERE in EMSAs, providing evidence of specific binding. Furthermore, competition binding with cold ERE and mERE confirmed that this interaction was specific to the wild-type response element (Fig. 3A). Purified recombinant Pin1 used in our assays is catalytically active and causes cis/trans-isomerization of the Ser(P) 118 -Pro 119 bond of purified ER␣, as determined by increased sensitivity to dephosphorylation by trans-specific PP2A (Fig. 3B) (25). FP using a DNA probe consisting of a consensus ERE was used to quantitatively determine the K d of ER␣ DNA binding in the presence and absence of Pin1. The consensus ERE used for the FP assays consisted of the core AGGTCA-CCG-TGACCT flanked by constant regions that formed a hairpin DNA. Pin1 significantly improved the K d of the ER-ERE interaction by at least 2-fold (4.0 Ϯ 1.6 nM versus 2.0 Ϯ 0.6 nM, p ϭ 0.003), and a representative experiment is shown in Fig. 3C. Similar to findings with EMSAs, there was undetectable binding of ER␣ to mERE in FP experiments under all conditions tested.
Dose-response curves indicated that Pin1 regulation of ER␣ DNA binding occurs within a broad range of Pin1 concentrations (Fig. 3, D and E). EMSA analysis indicated that Pin1 increased ER␣ binding affinity at doses ranging from 0.060 to 12 M, with maximal binding occurring at 1.5 M, and higher doses of Pin1 showed a diminution of ER␣-DNA binding. In . Phosphorylation by EGF is insufficient to enhance Pin1-mediated effects on ER␣-DNA binding. A, Western blot of ER␣ phosphorylated at Ser 118 in response to 100 ng/ml of EGF at the indicated time points. NS, nonspecific band for loading control. B, nuclear extracts (10 g) from MCF7 cells stably overexpressing GFP (vector (V)) or GFP-Pin1 (Pin1) treated with (EGF) and without (Ϫ) 100 ng/ml EGF were incubated with ERE followed by gel electrophoresis and autoradiography. The ER␣-ERE complex was competed with unlabeled ERE or mERE. SS, ER␣ supershift using anti-ER␣ antibody. ERE represents unbound free probe. Bands in the oval box were quantified. C, EMSA autoradiography (B) was quantified using a PhosphorImager, and data are represented as means Ϯ S.E. (error bars) for at least three independent experiments. No statistically significant differences were observed (p Ͼ 0.05).
addition, a reproducible but unexplainable decrease in binding occurred at 600 nM Pin1. In agreement with the EMSA data, quantitative determination by FP indicated that although the addition of Pin1 enhanced the K d of ER␣ binding in all cases, the effect was greater at lower doses of Pin1 (2 M) relative to high doses (10 or 20 M) (Fig. 3E). These results suggest that Pin1 activity on ER␣ is concentration-dependent, and based on our EMSA analysis, Pin1 exhibits a biphasic dose response (Fig. 3D).
Next, the requirement for isomerization by Pin1 on ER␣ DNA interactions was tested. To avoid potential saturation of binding and to best visualize the effect of Pin1, subsequent EMSA experiments used a suboptimal concentration of 100 nM purified ER␣ unless otherwise noted (Fig. 4A). Juglone is a specific pharmacological inhibitor of the catalytic activity of Pin1 in vitro (49). Quantification of the EMSA results showed that there was an ϳ16-fold increase in DNA binding activity of purified ER␣ due to Pin1. This effect was completely abolished in the presence of juglone (Fig. 4, B and C). These results were recapitulated by FP, where juglone consistently blocked the effect of Pin1 on ER␣ DNA binding affinity (Fig. 4, D and E). The role of Pin1-mediated cis-trans-isomerization in ER␣ DNA binding function was further confirmed using a catalytically inactive (K63A) mutant Pin1 protein (50). Compared with wildtype Pin1, the K63A Pin1 mutant did not increase ER␣ DNA binding activity by either EMSA or FP (Fig. 4, F and G). These data provide evidence that Pin1 isomerase activity is necessary for Pin1 to enhance ER␣ binding to its response element.
Specificity and Affinity Landscape of ER␣ DNA Binding Properties Regulated by Pin1-Because all of the DNA binding experiments in previous figures were performed with just two variations of the consensus EREs, we next wanted to unbiasedly explore the global effect of Pin1 on the DNA binding profile of ER␣. For this, we utilized the high throughput DNA microarray, referred to as the SNAP array, which displays 385,000 double-stranded DNA sequences in unique spatially resolved features (44,45). On the custom-designed SNAP array, duplex DNA probes were designed to contain 9 permuted bp, or 4 9 ERE variations, preceded by a constant ER␣ half-site. Each probe was additionally flanked by a constant 5-bp region on each end to form 25-bp B-form DNA after hairpinning (Table 2) (44,45). Binding of purified recombinant ER␣ to the DNA was detected by fluorescence, and the comprehensive DNA binding specificity and affinity of ER␣ to all sequences assayed were analyzed using motif-finding algorithms (51,52) and sequence specificity landscapes (SSLs; Fig. 5A) (44). Motif analysis of the top fluorescence intensity sequences indicated that ER␣ alone and ER␣ ϩ Pin1 bound to the consensus ERE (Fig. 5A, top). How-ever, when SSL analysis was employed to display the binding intensity from all 9-mer core sequences (underlined) using the dual core-motif AGGTCA-NNNTGACCT and AGGTCA-TGACCTNNN, Pin1-induced differences in ER␣ binding to DNA were revealed.
In SSLs, intensities for each DNA sequence from the array are plotted on a series of concentric rings, where each array sequence in the center ring contains a subsequence that perfectly matches the motif, sequences in the second ring contain a subsequence with one mismatch to the consensus, sequences in the third ring contain a subsequence with two mismatches to the consensus, and so forth. The height of the peak corresponds to intensity and is proportional to binding affinity (K a ). Results indicate that the affinity of ER␣ binding to DNA, in particular to DNA sequences bearing the consensus ERE AGGTCANNNT-GACCT (Fig. 5A), were enhanced in the presence of Pin1. Pin1 affected ER␣ binding to the AGGTCA presented within the context of a palindromic sequence (Groups D, I, and L as examples; Table 2 and Fig. 5B) but not with a simple AGGTCA halfsite (Groups R and FF; Table 2 and Fig. 5B). Moreover, DNA sequences bearing inverted repeats were preferred over direct repeats. These findings are observed in the SSLs, in which the sequences in the central ring (perfect match to the SSL consensus; shown by arrows) of the ER␣ ϩ Pin1 landscape are of significantly higher intensity than the central ring of the ER␣alone landscape. Comparison of the SSLs showed several Pin1-induced high intensity peaks in the outer rings, and these sequences correspond to AGGTCA direct repeats of 0 and 3 spacers, which are known ER␣ targets (53). A ridge of higher intensity peaks radiating outward in the ER␣ ϩ Pin1 landscape also contains sequences in which the motif degrades starting at the 3Ј-end. The first ring in the ridge contains NNNTGACCN, and the subsequent second ring includes NNNTGACNN. ER␣ plus Pin1 demonstrated preferences for the consensus ERE and 3-bp spacer, specifically the spacer CCG (labeled Seq1). Additional spacer analysis indicated that Pin1 also enhanced ER␣ interaction at inverted repeats with 0-, 2-, and 3-bp spacers.
To confirm the ER␣ signal enhancement on the SNAP array by Pin1, we used FP to quantify the effect of Pin1 on ER␣ binding to three different DNA sequences (Seq1, Seq2, and Seq3; 9-mer embedded within the 25-bp hairpin) (Fig. 5, C-E). As expected, the EREs bearing a consensus AGGTCANNNTGA-CCT resulted in the highest affinity of ER␣. A plot of K a versus normalized intensity yielded a linear correlation (r 2 Ͼ 0.94), thereby allowing the intensity of every 9-mer sequence to be related to affinity by K a (Fig. 5F). Consistent with our previous EMSA and FP results, the K a plot also shows that Pin1 enhances the affinity of ER␣ for the three sequences analyzed by FP.
FP and SNAP analysis indicated that Pin1 does not form a stable interaction with the ER␣-ERE complex but has a transient enzymatic interaction perhaps similar to other "allosteric cofactors" (54,55). The FP curve for ER␣ in the presence of Pin1 reached a lower plateau than ER␣ alone. FP provides readout on both the shape and size of the ER␣-DNA complex (56). Hence, the change in the ER␣ ϩ Pin1 FP curve is consistent with Pin1 inducing a conformational change in the ER␣ protein (25).
Pin1 Impacts ER␣ Recruitment to Endogenous Estrogen Response Elements-We used a ChIP assay in MCF7 cells overexpressing Pin1 or vector to probe for ER␣-DNA interaction at ER␣ binding sites on endogenous targets. Binding sites interrogated include the well characterized pS2 promoter, the ESR1 distal region (ENH1; Ϫ150 kb), the ESR1 promoter (ϩ60), and a nonspecific intervening region (Ϫ811) (46) (Fig. 6, A-C). ER␣ enrichment on ENH1 and pS2 ERE sites were higher in Pin1overexpressing cells (Fig. 6, A and B). In agreement with our SNAP data, these sites bear near perfect EREs, supporting the notion that Pin1 enhances ER␣ occupancy at the ERE sites most resembling a perfect palindromic consensus. In contrast, Pin1 had essentially no impact on the ER␣ binding site on the ESR1 proximal promoter region (ϩ60) (Fig. 6C) and a nonspecific region (Ϫ811) (Fig. 6, A-C). The latter two sites lack a consensus ERE, reflected by the low percentage of input DNA bound, although the ESR1 promoter is a site of estrogen-regulated ER␣ binding (46). Although the effect of Pin1 on ER␣ occupancy on DNA in cells is not as robust as in in vitro studies, the trend is consistent with EMSA and FP findings that Pin1 regulates ER␣ interaction with EREs on target genes.

Discussion
ER␣ and Pin1 are important regulatory molecules in cell division and growth control that have overlapping and distinct roles in cancer (57,58). The growth-promoting activity of ER␣ is mediated through the transcriptional regulation of gene expression. The mechanisms governing ER␣ transactivation are well described, and hundreds of ER␣ protein partners have been invoked as part of the process (59). They include multiple chromatin-remodeling and chromatin-modifying enzymes as well as the multiprotein Mediator complex, which bridges ER␣ with the basal transcriptional machinery. Phosphorylated ER␣ comprises ϳ48% of the total ER␣ in the presence of ligand (60), and gene expression analysis indicates that phosphorylated ER␣ regulates a subset of ER␣ target genes involved in cell growth (32). Pin1 was discovered as a critical regulator of mitosis and is overexpressed in a number of cancers, including breast, prostate, lung, and colon, among others relative to normal controls (27,57,61,62). A large number of substrates have been reported for Pin1, including critical cancer oncogenes (e.g. c-Myc, cyclin D1, and p53), transcriptional co-regulators (e.g. SMRT and AIB1), chromatin, and components of the transcriptional machinery, such as RNA polymerase II (30,63). Pin1 is also known to associate with several nuclear receptors, including glucorticoid receptor, PPAR␥, Nur77, androgen receptor, and retinoic acid receptor (38, 64 -66, 68). The broad activities of ER␣ and Pin1, coordinated through multiprotein complexes,

. Pin1 isomerase function increases ER␣ binding affinity to ERE.
A, to assess various levels of ER␣-ERE binding, recombinant purified ER␣ (0 -300 nM) was first incubated at 30°C for 30 min and then incubated with 32 P-labeled ERE, followed by gel electrophoresis and autoradiography. B, in vitro EMSA was performed as in (Fig. 3D)  make it difficult to dissect the specific activities of these partners in transcription and growth.
Our previous studies showed that Pin1 binds to the phosphorylated ER␣, and Pin1 isomerase activity increases receptor transactivation function in both the presence and absence of ligand (25). However, given the extensive protein interactions of both ER␣ and Pin1, an understanding of the direct role of Pin1 in ER␣ transactivation was lacking. In this study, we provide evidence that Pin1 can directly increase ER␣ DNA binding affinity, providing a plausible explanation for increased ER␣ transcriptional function by Pin1 in estrogen-treated cells. Using both breast cancer cell lysates and purified components, our data indicate that the overexpression or addition of Pin1 is sufficient to increase ER␣ affinity for DNA binding sites. Pin1 isomerase activity is responsible for the resultant effect, suggesting that Pin1-induced conformational changes are involved.
These studies, combined with our previous studies (25), support a model in which Pin1-catalyzed isomerization induces conformational changes in the ER␣ N terminus, which in turn modulates the ability of ER␣ to dimerize, recognize DNA binding sites, and regulate transcription. Order of addition experiments as well as inhibition of Pin1 catalytic function indicate that Pin1 isomerase activity must occur prior to the effects on ER␣-DNA binding affinity. These Pin1-mediated effects can act on unliganded ER␣ but are further enhanced for estrogenbound ER␣, which suggests that conformational changes from the N terminus are coordinated with ligand-induced changes in the LBD to modulate the activities of the DBD. Of interest, Pin1 enhances transactivation of ER␣ in the presence of OHT (25) but does not significantly increase the binding of OHT-occu-pied ER␣ to DNA. This finding points to context-specific control of ER␣ by Pin1 and indicates that Pin1 enhances tamoxifendriven transcriptional activity through a separate mechanism. In either case, our data reveal that Pin1 regulation of ER␣ transcription involves the integrated activities of multiple functional domains.
Interestingly, Pin1 does not directly interact with the ER␣ DBD but rather binds to a phosphorylation site in the adjacent ID region (25). In addition to Ser 118 , Pin1 has also been reported to bind to Ser 294 in the disordered hinge region immediately downstream of the DBD (69). Changes in DNA binding activity coordinated by ID regions have been observed in other transcription factors. For example, the ID regions in Cys 2 -His 2 zinc finger transcription factors lock the DNA binding motif into conformations that contribute to tighter and specific DNA interactions (70,71). Moreover, ID regions can also coordinate hydrophobic and electrostatic interactions that shape the interface of the DBD and DNA, as has been shown for Max and Ets factors (72,73). ID regions are mostly enriched with proline residues, and proline isomerization has been shown to kink the protein backbone which functions as a molecular switch to regulate the structure of nearby domains (74 -76). Therefore, allosteric regulation of the ER␣ DBD through extended conformational changes originating in the AF1 domain is plausible, although the lack of effect of EGF, which induces phosphorylation and Pin1 binding but does not affect ER␣ DNA binding, suggests additional complexities. Given the high proportion of ID regions in transcription factors that are common sites of phosphorylation, it is possible that the effects of Pin1 on DNA binding observed here on ER␣ may be more generally applied  Pin1 may also regulate ER␣ DNA binding via promoting interdomain interactions. As mentioned previously, the impor-tance of ligand in modulating Pin1 activity implicates coordinated regulation of the N-terminal AF1 and the C-terminal LBD on the DBD. This finding is in agreement with other results, which indicate that the N and C termini of receptors FIGURE 5. Effect of Pin1 on ER␣ DNA binding using the comprehensive SNAP microarray. A, top, motif for estrogen-occupied ER␣ binding to all permutations of the core sequence, AGGTCAN 9 , in the 25-bp hairpin on the SNAP array. Bottom, SSL for estrogen-occupied ER␣ binding to all 262,144 (9-mer) DNA sequences. Each ring contains sequences with increasing mismatches from the seed motif composed of the sequences NNNTGACCT and TGACCTNNN. Left, motif and SSL for ER␣. Right, motif and SSL for ER␣ ϩ Pin1. B, the effect of Pin1 was determined by plotting normalized fluorescence intensity for ER␣ (Array1_noPin1) versus ER␣ ϩ Pin1 (Array1_wtPin1) binding to different DNA probes from the groups listed in Table 2. A representative graph from two independent replicates is shown. Error bars, 1 S.D. C-E, determination of K d values for estrogen-occupied ER␣ (filled circles) or ER␣ ϩ Pin1 (open squares) for three DNA sequences (Seq1, Seq2, and Seq3; shown by arrows in Fig. 5A) from the SNAP array using FP. F, correlation between normalized SNAP intensity and K a , relating fluorescence intensities on the SNAP arrays to affinity of ER␣ (filled circles) or ER␣ ϩ Pin1 (open squares) for each DNA sequence. coordinately regulate receptor activity (77)(78)(79). Other receptor binding partners have been identified that enhance allosteric interactions among functional domains. Direct p300 interaction with the N-terminal A/B region of ER␣ not only bridges the N and C termini of ER␣ but also enhances receptor transactivation through ligand-induced synergism between AF1 and AF2 (80). Similarly, Dutertre and Smith (24) showed the importance of AF1 in recruiting SRC1 (steroid receptor coactivator-1) and CBP (CREB-binding protein) in mediating interaction and transcriptional synergism between AF1 and AF2 domains. Previous studies have also suggested that AF1 coregulator binding may have long range effects through changes in structure. Like Pin1, TBP binds to the N-terminal AF1 domains of ER␣ as well as the glucorticoid receptor and progesterone receptor (10,(81)(82)(83). In the case of glucorticoid receptor, TBP binding induces the transition to an ordered structure that facilitates interaction with the coactivator SRC1 to enhance ligand activated transcription. Recent studies of progesterone receptor using hydrogen-deuterium exchange have shown that TBP binding to progesterone receptor AF1 domain stabilized the AF2 domain, supporting the notion that coregulator binding at the AF1 domain may facilitate physical interactions between the AF1 and AF2 to alter transcriptional responses (83). NMR analysis of the ER␣ N terminus showed that, unlike TBP, Pin1 binding and isomerization of ER␣ does not induce an ordered structure in the N terminus (25). Pin1 did, however, increase ER␣ dimerization (25), which is necessary for receptor DNA binding function. We propose that Pin1 probably plays multiple roles in directing ER␣ DNA binding function by inducing conformational changes that are transmitted to both the DBD and dimer interface and that can be further modulated by ligand interactions in the LBD.
The SNAP-SSL analysis provided an unprecedented view into the effects of Pin1 on the ER␣-ERE interactions. For our analysis, we employed the SNAP technology which, unlike ChIP-sequencing, can distinguish direct ER␣-DNA interactions from indirect/protein-tethered interactions with DNA and also overcomes complications associated with Pin1 interactions with basal transcriptional machinery. Moreover, although ChIP-sequencing is an established and powerful technique, it requires next generation sequencing expertise, takes days or weeks to complete depending on the number of lanes needed, generates complicated data, is particular to a cell type or condition, is labor-intensive, and includes DNA amplification (84 -86). SNAP arrays, however, can be performed with purified proteins or lysates with high throughput. SNAP arrays were particularly well suited for our analysis because we wanted to pinpoint the specific biochemical effects of Pin1 on the ER␣-ERE interaction without the input of other cellular proteins. The DNA probes examined on the SNAP array contained the core sequence AGGTCA-9-mer embedded within a 25-bp hairpin, allowing for an unbiased analysis of the 3-mer spacer and 6-mer half site (4 9 or 262,144 permutations). Motif searching indicated an ER␣ preference for the consensus TGACCT at the second half-site in the presence or absence of Pin1.
SNAP-SSL analysis of all 9-mer permutations adjacent to the AGGTCA sequence further uncovered additional Pin1-induced effects on ER␣-DNA binding specificity and affinity. The most apparent effect of Pin1 was an enhancement of ER␣ affinity for the consensus ERE, but Pin1 caused ER␣ to be recruited to other ER␣ binding sites as well, including inverted repeats of zero, two, and three spacers as well as AGGTCA direct repeats. The SNAP data indicated that ERE spacer sizes of 0-and 3-mer are preferred (Groups T and W; Table 2 and Fig. 5B), as has been observed previously (53). In addition, enhanced ER␣ interaction at an inverted repeat is consistent with ER␣ dimer binding, and our previous findings indicated that Pin1 enhances ER␣ dimerization (25). Data showing that ER␣ ϩ Pin1 consistently preferred the CCG spacer within the context of the inverted repeat suggest that Pin1 may also alter ER␣-DNA interactions directly via subtle spacer sequence preferences.
To further support Pin1 effects on ER␣ specificity, we performed a differential binding analysis of the full 262,144 SNAP sequences. In comparing ER␣ versus ER␣ ϩ Pin1 affinities in this analysis, ER␣ ϩ Pin1 is enriched for strong palindromic ERE sequences, whereas ER␣ alone is enriched for sequences with a nearly unidentifiable degenerate second half-site. The ability of ER␣ to recognize a strong half-site with a highly degenerate second half-site has been observed in genome-wide ER␣ assays (67,87). Our comparative analysis of Pin1 enhancement of ER␣ binding at sites that best resemble the consensus ERE is further supported by our ChIP findings.
Identification of direct actions of Pin1 on ER␣ transactivation were facilitated by in vitro approaches, which revealed activities that would not be uncovered in the complex environments of the cell. The studies reported here indicate that although Pin1 may regulate multiple factors in the transcriptional process, at least one component of Pin1 isomerase activity on ER␣ is through the direct regulation of receptor DNA binding activity. Although Pin1 is not required for ER␣ binding to the DNA, these results suggest that one of the outcomes of Pin1 binding is to tune the receptor function and contribute to the specificity of gene regulation by phosphorylated forms of the receptor. Importantly, the identification of Pin1 as an ID region-binding protein that changes ER␣ interactions with DNA opens the door to future structural studies to pursue the allosteric transmission of signal from an important yet poorly understood ID region of the ER␣ to other functional domains of the receptor. Such insight would contribute to an understanding of how communication between ID functional domains of nuclear receptors and other Pin1-regulated transcription factors coordinate DNA binding functions and transcriptional activation, offering novel avenues to target and control Pin1 interactions in disease.