- Email: [email protected]
Molecular Dissection of the Intrinsically Disordered Estrogen Receptor Alpha-NTD
Structure
Previews Molecular Dissection of the Intrinsically Disordered Estrogen Receptor Alpha-NTD Karen G. Manalastas1 and Dmitri I. Svergun1,* 1European Molecular Biology Laboratory (EMBL), Hamburg Outstation, Notkestrasse 85, 22607 Hamburg, Germany *Correspondence: [email protected] https://doi.org/10.1016/j.str.2019.01.006
In this issue of Structure, Peng et al. (2019) use a combination of computational modeling and solution-based experimental methods to characterize the intrinsically disordered N-terminal transactivation domain of estrogen receptor alpha. A long-range contact (I33-S118) was identified and was demonstrated to affect the protein conformation and coactivator binding. Intrinsically disordered proteins (IDPs) are largely flexible macromolecules, which lack a defined three-dimensional organization. Despite the absence of rigid structure, IDPs are perfectly functional, and have been found to be abundant in eukaryotic genomes, where they are implicated in a wide range of functions. Many signaling proteins are IDPs or have intrinsically disordered regions (IDRs), and it is speculated that their inherent flexibility allows these proteins to bind promiscuously to other molecules in the signaling cascade (Wright and Dyson, 2015). Because of their flexibility, however, IDPs are extremely challenging targets for structure modeling. For one, the paradigm of describing proteins using a single high-resolution structure is largely inapplicable to IDPs, which must be represented as ensembles rather than single structures. Instead, the goal has been to collect parameters that describe this ensemble of states in terms of, for example, residual secondary structure content, long-range contacts, and regions of relative mobility (Eliezer, 2009). Each of these parameters is determined by a different biophysical characterization method, making IDPs a natural focus of integrative structure modeling. In this issue of Structure, Peng et al. (2019) use a combination of solutionbased methods and computational modeling to characterize the 187-residue long N-terminal transactivation domain of estrogen receptor alpha (ERa-NTD). ERa is a transcription factor implicated in the growth of most breast cancers (Rajbhandari et al., 2012). The ERa-NTD has previously been shown to bind the TATA-box binding protein (TBP), resulting in downstream transcription events €rnmark et al., 2001). In addition, (Wa
ERa-NTD phosphorylation on serine 118 was shown to enhance the recruitment of peptidyl prolyl isomerase Pin1, which in turn was demonstrated to increase proliferation potential in breast cancer cells in vivo (Rajbhandari et al., 2012). ERa-NTD is thus of great interest as a potential drug target; however, rational targeting of ERa-NTD requires further insights into its molecular features. Structural information about ERa-NTD was previously limited to circular dichroism (CD) data and chemical shifts from nuclear magnetic resonance (NMR) spectroscopy, which together showed that the NTD was mostly unstructured, with a high content of random coils (Rajbhandari €rnmark et al., 2001). et al., 2012; Wa Peng and co-workers confirmed these structural insights using the same methods, as well as with hydrogen-exchange mass spectrometry (HX-MS), which showed that nearly the entire length of the ERa-NTD backbone was exposed to solvent. HX-MS also unraveled some relatively more solvent-protected regions at residues 3–12 and 166–175, pointing to a possible presence of residual secondary structure (Peng et al., 2019). Small-angle X-ray scattering (SAXS) confirmed that the NTD was unstructured but more compact than expected for an IDP. Specifically, the expected radius of gyration (Rg) for an IDP of 187 residues was expected to be 39 A˚ based on the power law (Rgpred = 2.54 * N0.522, where N is the number of residues) (Bernado´ and Svergun, 2012), However the experimental Rg was significantly smaller, around 31 A˚. Upon the addition of a chemical denaturant to the ERa-NTD, the SAXS data reflected expansion of the structure, with an increased Rg = 43A˚, and a broader and
shifted distance distribution function, p(r), further indicating that the native ERa-NTD was more compact than a random coil. The next question to address was what caused this relative compactness. Yang and co-workers (Peng et al., 2019) combined SAXS data, hydroxyl radical footprinting (Takamoto and Chance, 2006) and all-atom molecular dynamics (MD) to answer this question. SAXS data was the source of size and shape information, while hydroxyl radical footprinting gave a measure of side chain solvent accessibility. Together, SAXS and the footprinting provide complementary information and this has previously been used to probe the structure of oligomers (Wang et al., 2011), protein complexes (Huang et al., 2016), and multidomain proteins (Huang et al., 2018). However, since ERa-NTD is an IDP, both SAXS and hydroxyl radical footprinting yielded information that described the ensemble average, rather than a single model. Thus, MD was employed to generate a large number of tentative ERa-NTD structures, from which a subset ensemble was selected to fit both the SAXS and footprinting data (Figure 1). This subset could be considered a good representation of the distribution of ERa-NTD conformations in solution (Peng et al., 2019). The selected subset ensembles were utilized to construct a consensus protein contact map. From this map, a long-range contact was found between two 5-residue stretches, each centered at residues I33 and S118. Peng et al. (2019) further studied this putative contact, given the previously demonstrated significance of S118 in breast cancer cell proliferation (Rajbhandari et al., 2012). From the hydroxyl radical footprinting, residue I33 was shown to be
Structure 27, February 5, 2019 ª 2019 Elsevier Ltd. 207
Structure
Previews
protected from solvent by a IDPs and specifically to find factor of two compared to drug targets in those of the adjacent residues, promedical importance. viding an indirect evidence ACKNOWLEDGMENTS that I33 may be involved in an intra-protein interaction. We acknowledge scientists working The I33-S118 interaction with IDPs whose work is left uncited was additionally analyzed by here due to space limitations. 19 NMR, utilizing F labeling of C118 by 3-bromo-1,1,1REFERENCES trifluroacetone (BTFA) to probe for solvent exposure. Bernado´, P., and Svergun, D.I. (2012). Structural analysis of intrinsiThree constructs were made cally disordered proteins by smalland compared: (1) NTDS118Cangle X-ray scattering. Mol. Biosyst. 8, 151–167. BTFA served as the most native variant, mutated only Eliezer, D. (2009). Biophysical characto accommodate the BTFA laterization of intrinsically disordered bel; (2) NTDS118C-BTFA-L31A proteins. Curr. Opin. Struct. Biol. 19, 23–30. as a negative control, because the L31A mutation was suffiHuang, W., Ravikumar, K.M., Parisien, ciently close to I33, but it M., and Yang, S. (2016). Theoretical was not expected to particimodeling of multiprotein complexes by iSPOT: Integration of small-angle pate in the long-range interacX-ray scattering, hydroxyl radical foottion; and (3) triple mutant printing, and computational docking. S118C J. Struct. Biol. 196, 340–349. -BTFA-K32A/I33A/ NTD P34A, for which the interacHuang, W., Peng, Y., Kiselar, J., tion was expected to be disZhao, X., Albaqami, A., Mendez, D., rupted. Only the triple mutant Chen, Y., Chakravarthy, S., Gupta, 19 S., Ralston, C., et al. (2018). resulted in a shift on the F Multidomain architecture of estrogen spectra, indicating that the receptor reveals interfacial cross-talk Figure 1. Ensemble Modeling of ERa-NTD interaction could be localized between its DNA-binding and ligandA pool of random models of ERa-NTD was generated using all-atom molecular binding domains. Nat. Commun. to this 3-residue window, as dynamics. Using a genetic algorithm, subsets from the pool were selected 9, 3520. the mutation outside this based on how well they fit the SAXS and hydroxyl radical footprinting data. The selected subsets were then used to derive protein contact information. Peng, Y., Cao, S., Kiselar, J., Xiao, window did not result in X., Du, Z., Hsieh, A., Ko, S., Chen, any spectra shift (Peng Y., Agrawal, P., Zheng, W., et al. et al., 2019). (2019). A metastable contact and The detected long-range contact was on the ERa-NTD as this phosphomimetic structural disorder in the estrogen receptor transactivation domain. Structure 27, this issue, then examined for functional implications mutation. Also, since it was previously es- 229–240. through a phosphomimetic mutation, tablished that S118 phosphorylation is S118D. The CD data revealed an increase important in regulating ERa-NTD function, Rajbhandari, P., Finn, G., Solodin, N.M., Singarapu, K.K., Sahu, S.C., Markley, J.L., in secondary structure upon this mutation. particularly through its interaction with Kadunc, K.J., Ellison-Zelski, S.J., Kariagina, A., The impact on the tertiary structure was Pin1, it would also be of interest to find Haslam, S.Z., et al. (2012). Regulation of estrogen receptor a N-terminus conformation and function determined through SAXS, which showed out how the S118D mutation affects by peptidyl prolyl isomerase Pin1. Mol. Cell. Biol. an Rg = 38.7 A˚, much larger than for the ERa-NTD-Pin1 binding. 32, 445–457. native protein, thus indicating a possible This study by Yang and co-workers Takamoto, K., and Chance, M.R. (2006). disruption of the I33–S118 contact. The (Peng et al., 2019) provides an excellent Radiolytic protein footprinting with mass specimpact of the mutation on TBP binding example of how various structure charac- trometry to probe the structure of macromolecular complexes. Annu. Rev. Biophys. Biomol. Struct. was also examined by surface plasmon terization methods can be combined, not 35, 251–276. resonance, yielding a weak-to-modest only to get a better picture of the IDP mobinding affinity of 53 ± 12 M between the lecular features, but also to gain further Wang, X., Watson, C., Sharp, J.S., Handel, T.M., and Prestegard, J.H. (2011). Oligomeric structure wild-type ERa-NTD and TBP, while the functional and mechanistic insights. While of the chemokine CCL5/RANTES from NMR, MS, phosphomimetic mutant had a more than the individual biophysical techniques and SAXS data. Structure 19, 1138–1148. 10-fold increase in binding affinity (Kd = used are widely utilized on their own, their Wa€rnmark, A., Wikstro¨m, A., Wright, A.P., 3.9 ± 0.4 M). Together, these data indicate combination was demonstrated to be Gustafsson, J.A˚., and Ha€rd, T. (2001). The that S118D mutation induces significant very powerful especially for IDPs. In partic- N-terminal regions of estrogen receptor a and b are unstructured in vitro and show different TBP binding conformational changes and strongly ular, the use of the structural ensembles to properties. J. Biol. Chem. 276, 45939–45944. affects coactivator binding. It would be derive a long-distance contact, and the Wright, P.E., and Dyson, H.J. (2015). Intrinsically interesting to further explore whether subsequent validation of this contact can disordered proteins in cellular signalling and reguphosphorylation induces the same effect be valuably applied to characterize other lation. Nat. Rev. Mol. Cell Biol. 16, 18–29. 208 Structure 27, February 5, 2019
Previews Molecular Dissection of the Intrinsically Disordered Estrogen Receptor Alpha-NTD Karen G. Manalastas1 and Dmitri I. Svergun1,* 1European Molecular Biology Laboratory (EMBL), Hamburg Outstation, Notkestrasse 85, 22607 Hamburg, Germany *Correspondence: [email protected] https://doi.org/10.1016/j.str.2019.01.006
In this issue of Structure, Peng et al. (2019) use a combination of computational modeling and solution-based experimental methods to characterize the intrinsically disordered N-terminal transactivation domain of estrogen receptor alpha. A long-range contact (I33-S118) was identified and was demonstrated to affect the protein conformation and coactivator binding. Intrinsically disordered proteins (IDPs) are largely flexible macromolecules, which lack a defined three-dimensional organization. Despite the absence of rigid structure, IDPs are perfectly functional, and have been found to be abundant in eukaryotic genomes, where they are implicated in a wide range of functions. Many signaling proteins are IDPs or have intrinsically disordered regions (IDRs), and it is speculated that their inherent flexibility allows these proteins to bind promiscuously to other molecules in the signaling cascade (Wright and Dyson, 2015). Because of their flexibility, however, IDPs are extremely challenging targets for structure modeling. For one, the paradigm of describing proteins using a single high-resolution structure is largely inapplicable to IDPs, which must be represented as ensembles rather than single structures. Instead, the goal has been to collect parameters that describe this ensemble of states in terms of, for example, residual secondary structure content, long-range contacts, and regions of relative mobility (Eliezer, 2009). Each of these parameters is determined by a different biophysical characterization method, making IDPs a natural focus of integrative structure modeling. In this issue of Structure, Peng et al. (2019) use a combination of solutionbased methods and computational modeling to characterize the 187-residue long N-terminal transactivation domain of estrogen receptor alpha (ERa-NTD). ERa is a transcription factor implicated in the growth of most breast cancers (Rajbhandari et al., 2012). The ERa-NTD has previously been shown to bind the TATA-box binding protein (TBP), resulting in downstream transcription events €rnmark et al., 2001). In addition, (Wa
ERa-NTD phosphorylation on serine 118 was shown to enhance the recruitment of peptidyl prolyl isomerase Pin1, which in turn was demonstrated to increase proliferation potential in breast cancer cells in vivo (Rajbhandari et al., 2012). ERa-NTD is thus of great interest as a potential drug target; however, rational targeting of ERa-NTD requires further insights into its molecular features. Structural information about ERa-NTD was previously limited to circular dichroism (CD) data and chemical shifts from nuclear magnetic resonance (NMR) spectroscopy, which together showed that the NTD was mostly unstructured, with a high content of random coils (Rajbhandari €rnmark et al., 2001). et al., 2012; Wa Peng and co-workers confirmed these structural insights using the same methods, as well as with hydrogen-exchange mass spectrometry (HX-MS), which showed that nearly the entire length of the ERa-NTD backbone was exposed to solvent. HX-MS also unraveled some relatively more solvent-protected regions at residues 3–12 and 166–175, pointing to a possible presence of residual secondary structure (Peng et al., 2019). Small-angle X-ray scattering (SAXS) confirmed that the NTD was unstructured but more compact than expected for an IDP. Specifically, the expected radius of gyration (Rg) for an IDP of 187 residues was expected to be 39 A˚ based on the power law (Rgpred = 2.54 * N0.522, where N is the number of residues) (Bernado´ and Svergun, 2012), However the experimental Rg was significantly smaller, around 31 A˚. Upon the addition of a chemical denaturant to the ERa-NTD, the SAXS data reflected expansion of the structure, with an increased Rg = 43A˚, and a broader and
shifted distance distribution function, p(r), further indicating that the native ERa-NTD was more compact than a random coil. The next question to address was what caused this relative compactness. Yang and co-workers (Peng et al., 2019) combined SAXS data, hydroxyl radical footprinting (Takamoto and Chance, 2006) and all-atom molecular dynamics (MD) to answer this question. SAXS data was the source of size and shape information, while hydroxyl radical footprinting gave a measure of side chain solvent accessibility. Together, SAXS and the footprinting provide complementary information and this has previously been used to probe the structure of oligomers (Wang et al., 2011), protein complexes (Huang et al., 2016), and multidomain proteins (Huang et al., 2018). However, since ERa-NTD is an IDP, both SAXS and hydroxyl radical footprinting yielded information that described the ensemble average, rather than a single model. Thus, MD was employed to generate a large number of tentative ERa-NTD structures, from which a subset ensemble was selected to fit both the SAXS and footprinting data (Figure 1). This subset could be considered a good representation of the distribution of ERa-NTD conformations in solution (Peng et al., 2019). The selected subset ensembles were utilized to construct a consensus protein contact map. From this map, a long-range contact was found between two 5-residue stretches, each centered at residues I33 and S118. Peng et al. (2019) further studied this putative contact, given the previously demonstrated significance of S118 in breast cancer cell proliferation (Rajbhandari et al., 2012). From the hydroxyl radical footprinting, residue I33 was shown to be
Structure 27, February 5, 2019 ª 2019 Elsevier Ltd. 207
Structure
Previews
protected from solvent by a IDPs and specifically to find factor of two compared to drug targets in those of the adjacent residues, promedical importance. viding an indirect evidence ACKNOWLEDGMENTS that I33 may be involved in an intra-protein interaction. We acknowledge scientists working The I33-S118 interaction with IDPs whose work is left uncited was additionally analyzed by here due to space limitations. 19 NMR, utilizing F labeling of C118 by 3-bromo-1,1,1REFERENCES trifluroacetone (BTFA) to probe for solvent exposure. Bernado´, P., and Svergun, D.I. (2012). Structural analysis of intrinsiThree constructs were made cally disordered proteins by smalland compared: (1) NTDS118Cangle X-ray scattering. Mol. Biosyst. 8, 151–167. BTFA served as the most native variant, mutated only Eliezer, D. (2009). Biophysical characto accommodate the BTFA laterization of intrinsically disordered bel; (2) NTDS118C-BTFA-L31A proteins. Curr. Opin. Struct. Biol. 19, 23–30. as a negative control, because the L31A mutation was suffiHuang, W., Ravikumar, K.M., Parisien, ciently close to I33, but it M., and Yang, S. (2016). Theoretical was not expected to particimodeling of multiprotein complexes by iSPOT: Integration of small-angle pate in the long-range interacX-ray scattering, hydroxyl radical foottion; and (3) triple mutant printing, and computational docking. S118C J. Struct. Biol. 196, 340–349. -BTFA-K32A/I33A/ NTD P34A, for which the interacHuang, W., Peng, Y., Kiselar, J., tion was expected to be disZhao, X., Albaqami, A., Mendez, D., rupted. Only the triple mutant Chen, Y., Chakravarthy, S., Gupta, 19 S., Ralston, C., et al. (2018). resulted in a shift on the F Multidomain architecture of estrogen spectra, indicating that the receptor reveals interfacial cross-talk Figure 1. Ensemble Modeling of ERa-NTD interaction could be localized between its DNA-binding and ligandA pool of random models of ERa-NTD was generated using all-atom molecular binding domains. Nat. Commun. to this 3-residue window, as dynamics. Using a genetic algorithm, subsets from the pool were selected 9, 3520. the mutation outside this based on how well they fit the SAXS and hydroxyl radical footprinting data. The selected subsets were then used to derive protein contact information. Peng, Y., Cao, S., Kiselar, J., Xiao, window did not result in X., Du, Z., Hsieh, A., Ko, S., Chen, any spectra shift (Peng Y., Agrawal, P., Zheng, W., et al. et al., 2019). (2019). A metastable contact and The detected long-range contact was on the ERa-NTD as this phosphomimetic structural disorder in the estrogen receptor transactivation domain. Structure 27, this issue, then examined for functional implications mutation. Also, since it was previously es- 229–240. through a phosphomimetic mutation, tablished that S118 phosphorylation is S118D. The CD data revealed an increase important in regulating ERa-NTD function, Rajbhandari, P., Finn, G., Solodin, N.M., Singarapu, K.K., Sahu, S.C., Markley, J.L., in secondary structure upon this mutation. particularly through its interaction with Kadunc, K.J., Ellison-Zelski, S.J., Kariagina, A., The impact on the tertiary structure was Pin1, it would also be of interest to find Haslam, S.Z., et al. (2012). Regulation of estrogen receptor a N-terminus conformation and function determined through SAXS, which showed out how the S118D mutation affects by peptidyl prolyl isomerase Pin1. Mol. Cell. Biol. an Rg = 38.7 A˚, much larger than for the ERa-NTD-Pin1 binding. 32, 445–457. native protein, thus indicating a possible This study by Yang and co-workers Takamoto, K., and Chance, M.R. (2006). disruption of the I33–S118 contact. The (Peng et al., 2019) provides an excellent Radiolytic protein footprinting with mass specimpact of the mutation on TBP binding example of how various structure charac- trometry to probe the structure of macromolecular complexes. Annu. Rev. Biophys. Biomol. Struct. was also examined by surface plasmon terization methods can be combined, not 35, 251–276. resonance, yielding a weak-to-modest only to get a better picture of the IDP mobinding affinity of 53 ± 12 M between the lecular features, but also to gain further Wang, X., Watson, C., Sharp, J.S., Handel, T.M., and Prestegard, J.H. (2011). Oligomeric structure wild-type ERa-NTD and TBP, while the functional and mechanistic insights. While of the chemokine CCL5/RANTES from NMR, MS, phosphomimetic mutant had a more than the individual biophysical techniques and SAXS data. Structure 19, 1138–1148. 10-fold increase in binding affinity (Kd = used are widely utilized on their own, their Wa€rnmark, A., Wikstro¨m, A., Wright, A.P., 3.9 ± 0.4 M). Together, these data indicate combination was demonstrated to be Gustafsson, J.A˚., and Ha€rd, T. (2001). The that S118D mutation induces significant very powerful especially for IDPs. In partic- N-terminal regions of estrogen receptor a and b are unstructured in vitro and show different TBP binding conformational changes and strongly ular, the use of the structural ensembles to properties. J. Biol. Chem. 276, 45939–45944. affects coactivator binding. It would be derive a long-distance contact, and the Wright, P.E., and Dyson, H.J. (2015). Intrinsically interesting to further explore whether subsequent validation of this contact can disordered proteins in cellular signalling and reguphosphorylation induces the same effect be valuably applied to characterize other lation. Nat. Rev. Mol. Cell Biol. 16, 18–29. 208 Structure 27, February 5, 2019