| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Directed evolution of estrogen receptor proteins with altered ligand-binding specificities
1Department of General and Developmental Physiology of Plants, Albrecht-von- Haller-Institute for Plant Sciences, Georg-August-University Goettingen, Untere Karspuele 2, 37073 Goettingen, Germany 2 Institute of Organic and Biomolecular Chemistry, Georg-August-University Goettingen, Tammannstr. 2, 37077 Goettingen, Germany 3Department of Chemical and Biomolecular Engineering, University of Illinois at Urbana-Champaign, 600 S Mathews Avenue, Urbana, IL 61801, USA
4 To whom correspondence should be addressed. E-mail: cgatz{at}uni-goettingen.de or cgatz{at}gwdg.de
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResults and discussionFundingAcknowledgementsReferences |
|---|
Transcriptional activators that respond to ligands with no cellular targets are powerful tools that can confer regulated expression of a transgene in almost all biological systems. In this study, we altered the ligand-binding specificity of the human estrogen receptor
(hER) so that it would recognize a non-steroidal synthetic compound with structural similarities to the phytoestrogen resveratrol. For this purpose, we performed iterative rounds of site-specific saturation mutagenesis of a fixed set of ligand-contacting residues and subsequent random mutagenesis of the entire ligand-binding domain. Selection of the receptor mutants and quantification of the interaction were carried out by exploiting a yeast two-hybrid system that reports the ligand-dependent interaction between hER and steroid receptor coactivator-1 (SRC-1). The screen was performed with a synthetic ligand (CV3320) that promoted growth of the reporter yeast strain to half maximal levels at a concentration of 3.7 µM. The optimized receptor mutant (L384F/L387M/Y537S) showed a 67-fold increased activity to the synthetic ligand CV3320 (half maximal yeast growth at 0.055 µM) and a 10-fold decreased activity to 17ß-estradiol (E2; half maximal yeast growth at 4 nM). The novel receptor-ligand pair partially fulfills the requirements for a specific ‘gene switch’ as it responds to concentrations of the synthetic ligand which do not activate the wildtype receptor. Due to its residual responsiveness to E2 at concentrations (4 nM) that might occur in vivo, further improvements have to be performed to render the system applicable in organisms with endogenous E2 synthesis.
Keywords: estrogen receptor/ligand specificity/saturation mutagenesis
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResults and discussionFundingAcknowledgementsReferences |
|---|
The mammalian steroid receptors are ligand-dependent transcriptional activators that have been engineered to serve as specific ‘gene switches’ in heterologous systems such as yeasts, plants and mammals (Louvion et al., 1993
; Aoyama and Chua, 1997; Gallinari et al., 2005). The basic strategy involves fusing the ligand-binding domain (LBD) of the receptor to a heterologous transcriptional activator which binds to a specific target promoter that is not recognized by endogenous transcription factors. A typical example is the chimeric transcriptional activator GVG that consists of the DNA-binding domain of yeast transcription factor GAL4, the activation domain of Herpes simplex virus protein VP16 and the LBD of the rat glucocorticoid receptor (Aoyama and Chua, 1997). This protein activates transcription from a target promoter containing GAL4 upstream activating sequences and a core promoter only in the presence of dexamethasone and related steroids and has been used for studying a number of biological processes in transgenic plants (Moore et al., 2006). In mammalian cells, such an activator would respond to endogenous steroids, and—vice versa—endogenous genes would respond to the steroid applied to activate the transgene. Therefore, numerous studies have been initiated to alter the ligand specificity of various receptors so that it would bind to compounds with no cellular targets (Chockalingam et al., 2005; Gallinari et al., 2005). This concept is also valuable for agricultural purposes as side effects on other organisms could be avoided.
The estrogen receptor is particularly suited to design such a so-called functionally orthogonal gene switch because it can accommodate a variety of non-steroidal compounds in its binding pocket (Gallinari et al., 2005). A minimal requirement is the distance between two terminal polar oxygen groups of the inducing molecule (Fig. 1 (Anstead et al., 1997; Brzozowski et al., 1997)) and an aromatic ring in the ‘A’ position, while the remaining part of the molecule can be quite different in structure. Indeed, more spacious chemicals like the antagonist tamoxifen or the agonists of the tetrahydrofluorenone-type fit into the pocket (Brzozowski et al., 1997; Gallinari et al., 2005).
|
Recently, a general method for the generation of specific receptor-ligand pairs was described (Chockalingam et al., 2005
). It involves the random replacement of each ligand-contacting residue by all possible 20 amino acids. The generated receptor mutants are selected for binding to a synthetic ligand using a high throughput yeast screening system. Subsequently, the site saturation mutagenesis is repeated with the resulting mutant. The ligand specificity can finally be enhanced by random point mutagenesis of the LBD and phenotypic screening for further improved mutants. Using the synthetic non-steroidal compound 4,4'-dihydroxybenzil (DHB), a mutant human estrogen receptor (hER) that showed a 50-fold enhanced binding activity to DHB and a 140-fold reduced affinity to its natural ligand E2 was selected after two rounds of mutagenesis. After three further rounds of mutagenesis and selection, a mutant receptor was identified that showed a 5-fold enhanced binding activity to DHB and an over 106-fold reduced affinity to E2 when compared with the wildtype protein. Thus, the selectivity was enhanced by over 107-fold.
Here we have applied this strategy to a class of synthetic non-steroidal compounds that are characterized by one aromatic ring with a hydroxyl group being connected to a substituted cyclohexenone via a C–C triple bond (Fig. 1B). The starting apparent affinity of the ligand to the wildtype receptor as determined by its effective concentration to promote half maximal growth (EC50) of the yeast reporter strain (Chockalingam et al., 2005) was 3.7 µM and was enhanced to 0.055 µM after three rounds of mutagenesis and selection. Conversely, the affinity of the mutant receptor to E2 was reduced by a factor of 10.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResults and discussionFundingAcknowledgementsReferences |
|---|
Synthesis of ligands
Estrogen analogs were synthesized using a Pd-catalyzed Sonogashira-reaction of an iodoarene derivative with propargylic alcohol followed by an oxidative elimination of the CH2OH-group and a second Pd-catalyzed Sonogashira reaction with a 3-iodocyclopentenone or a 3-iodocyclohexenone (Tietze et al., 2008). A broad range of substitution patterns was covered and a complete list of the compounds tested in this work is depicted in Supplementary data available at PEDS online, Figure S1.
Mutagenesis of the human estrogen receptor, selection of mutants, ligand dose-response assays and molecular modeling
A yeast two-hybrid system that reports the ligand-dependent interaction between hER and human SRC-1 was used for the selection of hER mutants and the dose-response assay (Chockalingam et al., 2005). Briefly, the bait GBD-hER contains most of the LBD (amino acids 303–553) and the F domain (amino acids 554–595) of hER fused to the GAL4 DNA-binding domain (GBD). SRC-1 fused to the GAL4 activation domain (GAD) served as a prey (AD-SRC-1). The GBD-hER/AD-SRC-1 complex, which is formed only after binding of agonist ligands to LBD, activates the selection marker gene HIS3 enabling the yeast strain YRG2 to grow on histidine drop-out medium. Plasmids, mutant library creation and screening, ligand dose-response assays and molecular modeling were carried out as described previously (Chockalingam et al., 2005). Molecular graphics images were produced using the UCSF Chimera package from the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco (supported by NIH P41 RR-01081) (Pettersen et al., 2004).
Plasmids used in the yeast one-hybrid system
Plasmid pGBT-GEV is a pGBT9 (Clontech Laboratories Inc., Saint-Germain-en-Laye, France) derivative that contains the GBD (aa 1–92), the human estrogen receptor ligand-binding domain (hER LBD; aa 282–595) and the VP16 activation domain (aa 423–490) as in pHCA/Gal4ERVP16 (Louvion et al., 1993) behind the ADH1 promoter. Derivatives of pGBT-GEV containing the wildtype or engineered LBDs were transformed into yeast strain PJ69-4A according to Gietz and Woods (Gietz and Woods, 2002). Quantitative β-galactosidase assays were performed as described (James et al., 1996).
|
Results and discussion |
|---|
|
TopAbstractIntroductionMethodsResults and discussionFundingAcknowledgementsReferences |
|---|
Rationale for the synthesis of suitable ligands
Estrogen receptor agonists induce a specific conformational switch of cognate receptor proteins, which lead to the dissociation from the inhibitory protein HSP90 (Yamamoto et al., 1988) and allow the association with the co-activator SRC-1 (Onate et al., 1995). Therefore, it can be envisioned that the starting ligand for a strategy designed to obtain receptor proteins with altered ligand specificities should already weakly promote activation of hER. Typically, estrogen receptor ligands comprise two hydroxyl groups separated by a rigid hydrophobic linker region consisting of four carbocyclic rings. The hydroxyl group at the ‘A’ ring forms hydrogen bonds with the polar amino acids arginine 394 (R394) and glutamic acid 353 (E353) of hER, whereas the hydroxyl group at the D ring forms a hydrogen bond with histidine 524. This arrangement is important for the ligand-induced conformational change (Brzozowski et al., 1997). Therefore, we designed chemicals (Fig. 1B and Supplementary data are available at PEDS online, Fig. S1) that contain two polar oxygen containing groups at a distance of 10–12 Å which is in the same range as the distance of the oxygen atoms in the estradiol (E2) molecule. The two aromatic rings of the synthetic compounds which correspond to the ‘A’ and ‘D’ rings of the E2 molecule are connected either by a linear two-carbon single, double or triple bond spacer rather than by the carbocyclic rings found in E2. The scaffold of the ligands has similarity to the phytoestrogen resveratrol found in red wine (Fig. 1C), which activates hER at concentrations of 3–10 µM (Gehm et al., 1997). As phytoestrogens are phytoalexins with anti-fungal activity (Hain et al., 1990), they do not necessarily have plant and mammalian targets when used at low doses, though beneficial activity through the interaction with sirtuins has been observed for these compounds (Howitz et al., 2003). The aromatic ‘A*’ ring of the synthetic ligands carries the hydroxyl group in a position analogous to the hydroxyl group in E2. The ‘D*’ ring being either a 5- or a 6-membered ring carries a keto-group similar to the E2 homologue estron which renders the synthetic compounds more stable. The absence of the ‘B’ and the ‘C’ ring should lead to a reduced affinity to the wildtype receptor and the introduction of additional substituents to rings ‘A*’ or ‘D*’ should provide contact sites to mutant receptor proteins.
Selection of appropriate ligands for the creation of an orthogonal ligand-receptor pair
Twenty-seven different ligands of our compound library were grouped into three classes according to their capacity to promote yeast growth by facilitating the interaction between the wildtype receptor LBD and its co-activator SRC-1 in the yeast two-hybrid system (A: 0.01–0.09 µM; B: 0.1–0.9 µM; C: 1–9 µM; Supplementary data are available at PEDS online, Fig. S1). All experiments were performed in 96-well plates that allow measuring cell density at OD600. Cells expressing both the bait (GBD-hER) and the prey protein (AD-SRC) showed the half-maximal growth at an effective concentration (EC50) of 0.42 nM E2, which is in good agreement with the reported value for ligand-dependent transactivation by the full-length hER in yeast and CHO cells (Wrenn and Katzenellenbogen, 1993). Class A contains only one compound (CV5407 (Fig. 2A)) which shows an EC50 of 0.05 µM (Fig. 2B). Exchanging the chlorine of CV5407 to a methyl group decreased the EC50 by a factor of 10 (0.5 µM; Fig. 2B). Addition of a further methyl group in R2 generated a compound with an EC50 value of 4 µM (class C ligand; Fig. 2A and B). Like resveratrol (Fig. 1C; Gehm et al., 1997), compounds containing a C2-double bond or a C2-single bond (R3) had high EC50 values of 6 µM (Supplementary data are available at PEDS online, Fig. S1). As expected, protection of the phenolic hydroxyl group at C-3 essentially abolished the interaction (Fig. 2A and B).
|
Classification of the ligands according to their affinities to the wildtype receptor was confirmed in the yeast one-hybrid assay. This system is based on a chimeric transcriptional activator (GBD-hER
-VP16) consisting of the yeast GBD, the hER ligand-binding domain (LBD) (aa 282–576) and the VP16 activation domain. The yeast strain PJ69 encodes the ß-galactosidase reporter genes under control of a target promoter containing GAL4 upstream activating sequences (GAL7:lacZ) (Louvion et al., 1993). Activation of this reporter gene by GBD-hER-VP16 depends on the binding of E2 to LBD, which leads to the dissociation of the protein from the inhibitory protein HSP90, allowing translocation into the nucleus (Yamamoto et al., 1988). As shown in Fig. 2C, the relative effective concentrations of the various ligands established with the two-hybrid system were reproduced. However, 20% of the maximal response was obtained at 10 µM CV6019 with the yeast one-hybrid system, whereas 0.2 µM were sufficient in the yeast two-hybrid system, indicating that the one-hybrid system is approximately 200-fold less sensitive than the two-hybrid system. This can be explained by the fact that basal amounts of GBD-hER and GBD-hER-VP16 are in the nucleus even in the absence of the ligand, which in turn leads to higher background activities and thus to a reduced sensitivity of the GBD-hER-VP16-based system.
We performed our directed evolution studies with four group C ligands (Fig. 3A), thus reducing the risk of activating wildtype hER in the case of a future application. All four compounds activated the wildtype hER LBD to half maximum activity at concentrations between 2 and 7 µM in the yeast two-hybrid system, but did not increase background-level activity of the yeast one-hybrid strain even at a 10 µM concentration (Fig. 3C).
|
Saturation mutagenesis
The saturation mutagenesis strategy involves the replacement of each amino acid residue that is expected to contact the ligand with the 19 possible alternatives followed by the selection using the yeast two-hybrid system. To identify residues likely to interact with our synthetic ligands, we exploited the known co-crystal structure of hER LBD complexed with anti-estrogen diethylstilbestrol (DES) for molecular modeling studies (Shiau et al., 1998) (Fig. 4). Docking of class C ligand CV3320 into the binding pocket of hER LBD was performed using Molecular Operating Environment (MOE; Chemical Computing Group, Montreal, Canada) as described before (Chockalingam et al., 2005). Twenty-one residues that are in direct contact (within 4.6 Å) were identified. Three of the 21 residues, namely E353, R394 and H524, were excluded from individual site saturation mutagenesis because of their known essential role in hydrogen bonding with the terminal polar oxygen atoms of E2 (Brzozowski et al., 1997).
|
Each yeast library of receptor variants mutated at one of the 18 positions was plated onto minimal media agar plates that allowed selecting for both the prey and the bait plasmid. One hundred and ninety yeast colonies from each library were picked and subjected to the yeast two-hybrid screening by monitoring the growth on selective medium in a microtiter plate. The concentration of the synthetic ligands (CV2105, CV3320, CV6017 and IK524) was 0.5 µM, which is about 10-fold lower than the EC50 value determined with the wildtype receptor. Yeast cells expressing the non-mutagenized wildtype construct were included as controls in the screening step. Mutants were considered as promising candidates when yeast cells grew more vigorously when compared with the controls in the presence of the synthetic ligand and when the same mutant grew less vigorously in the presence of 0.5 nM E2.
In the first round of saturation mutagenesis, six putative candidate clones were identified with ligand CV3320 and two with ligand IK524. Sequence analysis revealed that the leucine residue at position 384 was exchanged to phenylalanine (L384F) in all eight clones. The clone showed a 10-fold improved activity towards CV3320, and a 90-fold weakened response towards E2 (Fig. 5A; Table I). The mutant protein also showed a 10-fold higher sensitivity towards IK524, but the sensitivity to CV6017 or CV2105 was not altered (data not shown). Next, the DNA of the mutant was taken as a template for performing further saturation mutagenesis in the remaining 17 positions. No clones with improved sensitivity towards CV3320 or IK534 could be obtained, but one of the mutants (L384F/L387M) revealed 8.6-fold reduced activity in combination with E2 (Fig. 5B; Table I). This double mutant also showed reduced sensitivity towards IK524 (data not shown). When performing a third round of saturation mutagenesis in the remaining 16 positions and a subsequent screen on CV3320, no further improvement was achieved.
|
|
Random mutagenesis
To improve the ligand-binding characteristics of the double mutant, error-prone PCR was performed to introduce random mutations throughout the entire LBD. Approximately 2 x 106 transformants were directly plated on selective medium containing 0.05 µM of ligand CV3320. Three hundred and seventy-eight clones were transferred into four 96-well master plates (minimal media selecting for the two plasmids) and incubated overnight at 30°C. Subsequently, colonies were inoculated into 96-well plates containing selective conditions for the two plasmids and for the transactivation (histidine drop-out medium). Four colonies that did not grow in the absence of the ligand and grew more vigorously than the original double mutant in the presence of 0.05 µM CV3320 were analyzed further. After plasmid rescue and retransformation, one clone showed a 5.8-fold increased affinity towards CV3320 compared to the L384F/L387M double mutant. This clone showed an 80-fold increased affinity towards E2 compared to the L384F/L387M double mutant (Fig. 5C; Table I). DNA sequencing revealed that tyrosine 537 was mutated into a serine (Y537S). The triple mutant (L384F/L387M/Y537S) finally served as a template for a second round of random mutagenesis. However, screening of a library of 3.2 x 106 variants did not result in a yeast colony with improved affinity towards CV3320 or decreased affinity towards E2. Table I summarizes the results of the three successful rounds of mutagenesis as determined with the yeast two-hybrid system.
In the yeast one-hybrid system, the L384F/L387M/Y537S triple mutant receptor in combination with 10 µM CV3320 conferred 
77% of the effect elicited by E2 in combination with the wildtype receptor. In contrast, the improved features of the L384F/L387M double mutant (EC50 value of 0.32 µM) were not detected (Fig. 5D). This was unexpected, because the interaction of the L384F/L387M double mutant to E2 (EC50 value of 0.32 µM) was detectable in this system. It could well be that the CV3320-induced conformational switch of GBD-hER has to be stabilized by the subsequent association with SRC-1, whereas 0.32 µM concentrations of this ligand are not sufficient for the efficient dissociation from HSP90. Thus, caution is required if a system that has been selected for improved ligand specificity with SRC-1 is applied in a context requiring dissociation from HSP90.
Accommodation of the synthetic ligand into the mutated receptor ligand-binding pocket
Though the size of the cavity (490 Å3) of hER is nearly twice the volume of E2 (245 Å3), we chose less space-filling ligands of the resveratrol type for our approach. Our rationale was to decrease the affinity of hER to the ligand by replacing rings ‘B’ and ‘C’ by an aliphatic spacer and to complement for the missing interactions by adding functional groups to ring ‘A*’ that would be recognized only by a mutated receptor protein. This strategy allowed us to test a compound library characterized by a common scaffold and variable substitutions. Indeed, one ligand (CV3320) with two methyl groups added to the ‘A*’ ring turned out to be a suitable candidate for the development of an orthogonal ligand-receptor pair. The saturation mutagenesis of amino acids lining the ligand-binding pocket (LBP) led to an exchange of L384 into F384 that caused increased affinity to CV3320. Figure 6 shows the relative position of this amino acid to CV3320 and E2. Due to the relative proximity of F384 to the methyl group of CV3320 (2 Å), a transient asymmetry in the electronic charge might be created, leading to an increased van der Waal's interaction. In contrast, the interaction with E2 is weakened, presumably due to space restrictions. Position 384 has been shown before to be involved in ligand specificity: hERβ, which differs from hER by recognizing a different subset of so-called ‘selective estrogen receptor modulators’ (SERMs) encodes a methionine at this position (Hillisch et al., 2004). Site-directed mutagenesis and transactivation studies suggested that differences in the ligand specificities of hER and hERβ are due to this amino acid exchange when occurring in combination with other mutations (Kumar et al., 2004).
|
The second round of mutagenesis did not yield a mutant with increased binding affinity to CV3320, indicating that no other favorable contacts to the ligand could be established within the LBP. However, the L387M mutation decreased the affinity to E2 and IK524, most likely due to space restrictions generated by the longer side chain of methionine when compared with leucine. Indeed, L387 contributes to the subpocket that accommodates the ‘A’ ring of E2 (Anstead et al., 1997
). The affinity to CV3320 is not reduced presumably because of the small size of the ligand. Further improvements of the ligand-binding selectivity for CV3320 could not be obtained using the stepwise saturation mutagenesis strategy.
Random mutagenesis of the LBD with the two point mutations L384F and L387M by error prone PCR yielded a mutant protein with a tyrosine to serine exchange at residue position 537, which is located 9 Å away from the ligand. The triple mutant (L384F/L387M/Y537S) showed increased affinity to CV3320 and E2 when compared with the double mutant. It has been reported that mutations of Y537 to N, A and S led to constitutive activation (Zhang et al., 1997; Yudt et al., 1999), which contrasts with our findings that activation of the triple mutant was still ligand-dependent. In our hands, the Y537S single mutant also did not lead to constitutive activation (Fig. 5D). It could well be that, when being part of a hybrid protein, this mutant does not flip into the active conformation though the conformational change seems to be facilitated upon binding of ligands.
In conclusion, we have shown that ligands containing rings corresponding to the A and the D ring of the steroid E2, which are connected by a C–C triple bond linker rather than by a rigid ring system, can be used as a scaffold for the development of a functionally orthogonal gene switch. The novel receptor-ligand pair responds to concentrations of the synthetic ligand (0.05 µM) which do not activate the wildtype receptor minimizing pleiotropic effects. Likewise, the commercially available RheoSwitch (New England Biolabs Inc.), which is based on the ecdysone receptor, uses ligands at concentrations ranging from 0.02–0.1 µM for optimum induction of target genes (Palli et al., 2003; Kumar et al., 2004).
Testing the system in mammalian or plant cells is necessary to further explore its potential applicability. As the affinity of the mutant hER receptor to E2 is only decreased by a factor of 10, one problem in mammalian cells might be the potential activation by endogenous E2 or other steroids. Likewise, in plants which lack endogenous E2 synthesis but—depending on the plant family—produce phytoestrogens (Hain et al., 1990), pre-induction might occur. As phytoestrogens also interact with chromatin modifying proteins of the sirtuin family (Howitz et al., 2003), it has to be tested whether CV3320 or related compounds affect sirtuin function at the concentrations used. Also, other questions like the half-life and mobility of the chemical in the foreign organism have to be addressed. Still, the structural motif of our non-steroidal ligands, i.e. the substitution of rings B and C of the original steroid framework by an alkine linker, has the potential to be applied in the future.
|
Funding |
|---|
|
TopAbstractIntroductionMethodsResults and discussionFundingAcknowledgementsReferences |
|---|
SFB ‘Chemical and Biological Synthesis and Transformation of Natural Products and Analogues’ of the Deutsche Forschungsgemeinschaft (DFG to K.M.D.I., I.K.K. and C.V.). National Science Foundation CAREER Award (BES-0348107 to H.Z.).
|
Footnotes |
|---|
Edited by Arne Skerra
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResults and discussionFundingAcknowledgementsReferences |
|---|
We thank J.A. Katzenellenbogen (University of Illinois at Urbana-Champaign) for his helpful comments.
|
References |
|---|
|
TopAbstractIntroductionMethodsResults and discussionFundingAcknowledgementsReferences |
|---|
Anstead G.M., Carlson K.E., Katzenellenbogen J.A. Steroids (1997) 62:268–303.[CrossRef][Web of Science][Medline]
Aoyama T., Chua N.H. Plant J. (1997) 11:605–612.[CrossRef][Web of Science][Medline]
Brzozowski A.M., Pike A.C., Dauter Z., Hubbard R.E., Bonn T., Engstrom O., Ohman L., Greene G.L., Gustafsson J.A., Carlquist M. Nature (1997) 389:753–758.[CrossRef][Medline]
Chockalingam K., Chen Z., Katzenellenbogen J.A., Zhao H. Proc. Natl Acad. Sci. USA (2005) 102:5691–5696.
Gallinari P., et al. Chem. Biol. (2005) 12:883–893.[CrossRef][Web of Science][Medline]
Gehm B.D., McAndrews J.M., Chien P.-Y., Jameson J.L. Proc. Natl Acad. Sci. USA (1997) 94:14138–14143.
Gietz R.D., Woods R.A. Methods Enzymol. (2002) 350:87–96.[CrossRef][Web of Science][Medline]
Hain R., Bieseler B., Kindl H., Schroder G., Stocker R. Plant Mol. Biol. (1990) 15:325–335.[CrossRef][Web of Science][Medline]
Hillisch A., Peters O., Kosemund D., Muller G., Walter A., Schneider B., Reddersen G., Elger W., Fritzemeier K.H. Mol. Endocrinol. (2004) 18:1599–1609.
Howitz K.T., et al. Nature (2003) 6954:191–196.
James P., Halladay J., Craig E.A. Genetics (1996) 144:1425–1436.[Abstract]
Kumar M.B., et al. J. Biol. Chem. (2004) 279:27211–27218.
Louvion J.F., Havaux-Copf B., Picard D. Gene (1993) 131:129–134.[CrossRef][Web of Science][Medline]
Moore I., Samalova M., Kurup S. Plant J. (2006) 45:651–683.[CrossRef][Web of Science][Medline]
Onate S.A., Tsai S.Y., Tsai M.J., O'Malley B.W. Science (1995) 270:1354–1357.
Palli S.R., Kapitskaya M.Z., Kumar M.B., Cress D.E. Eur. J. Biochem. (2003) 270:1308–1315.[Web of Science][Medline]
Pettersen E.F., Goddard T.D., Huang C.C., Couch G.S., Greenblatt D.M., Meng E.C., Ferrin T.E. J. Comput. Chem. (2004) 25:1605–1612.[CrossRef][Web of Science][Medline]
Shiau A.K., Barstad D., Loria P.M., Cheng L., Kushner P.J., Agard D.A., Greene G.L. Cell (1998) 95:927–937.[CrossRef][Web of Science][Medline]
Tietze L.F., Vock C.A., Krimmelbein I.K., Wiegand J.M., Nacke L., Ramachandar T., Islam K.M.D., Gatz C. Chem. Eur. J. (2008) 14:3670–3679.[CrossRef]
Wrenn C.K., Katzenellenbogen B.S. J. Biol. Chem. (1993) 268:24089–24098.
Yamamoto K.R., Godowski P.J., Picard D. Cold Spring Harbor Symp. Quant. Biol. (1988) 53:803–811.
Yudt M.R., Vorojeikina D., Zhong L., Skafar D.F., Sasson S., Gasiewicz T.A., Notides A.C. Biochemistry (1999) 38:14146–14156.[CrossRef][Web of Science][Medline]
Zhang Q.X., Borg A., Wolf D.M., Oesterreich S., Fuqua S.A. Cancer Res. (1997) 57:1244–1249.
Received July 2, 2008; revised September 18, 2008; accepted October 14, 2008.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||