Final Report Summary - HSP90NMR (The study of Hsp90 through the use of nuclear magnetic resonance, small angle scattering and calorimetry.)
Publishable Summary Report
Proper protein folding is essential for cellular homeostasis. Cells employ a large array of chaperones to ensure protein adopt the proper conformation. Hsp90 is an important chaperone and comprises of 1-2% of all protein is eukaryotic cells. Additionally, Hsp90 is upregulated in cancerous cells to facilitate fast proliferation. There are several X-ray crystal structures available for Hsp90 free in solution and interacting with its target proteins. This combined with small angle x-ray/neutron scattering (SAXS/SANS) and Förster resonance energy transfer (FRET) data, demonstrate that HSP90 is a 90 kDa dimeric protein and undergoes large conformational changes throughout its function. However, these crystal structures only provide a static picture of Hsp90 function so the molecular and dynamic details of these rearrangements are not well understood. We employ nuclear magnetic resonance spectroscopy (NMR) as a method to further probe these changes. NMR can provide atomically resolved real time information of Hsp90 in solution either free or interacting with its targets. Hsp90 contains three distinct domains, and we have individually isotopically labeled each domain. This allows each domain to be studied either independently or and in tandem with neighbouring domains. Also a target protein, the glucocorticoid receptor (GR) protein, has been introduced to observe how the different domains are affected by binding to a client protein. Through observing the changes in chemical shifts we have identified specific regions in Hsp90 which are involved in the binding of client proteins. We also employed SAXS/SANS experiments to observe changes in orientation of the domains when interacting with the GR protein. Through combining all of these techniques we have gained a broader understanding of Hsp90 function
We recorded and compared NMR spectra of full length, N-terminal domain (NTD), middle domain (MD), C-terminal domain (CTD) and NTD-MD tandem domains. We optimized a protocol to ligate together two domains using the Sortase Ligation Protein. This procedure allows us to specifically isotopically label individual domains in a multi domain construct and observe it through NMR. The final product contains two domains, one which we can see via NMR and one which is invisible. This greatly reduces spectral overlap so we are better able to analyze the behavior of each domain in a tandem complex, figure 1A. Through comparing the spectral changes between the individual domains and the ligated domain we identified a previously unknown region on NTD which interacts with charged linker region, figure 1B. The overall effects of this interaction on Hsp90 function is still unknown, but studies continue to identify possible effects.
The improved protocol is useful for a wide range of additional biophysical experiments, such as small-angle neutron scattering, electron paramagnetic resonance or fluorescence-based methods. Reducing the number of isotope-labeled amino acids by segmental labeling greatly simplifies NMR spectra. This can allow for the residue-specific analysis of NMR parameters, which may be obscured by signal overlap in a uniformly isotope-labeled sample. The method will be most useful when working with large multi-domain systems, i.e. where flexible linkers connect structurally independent domains.
Studying full length Hsp90 by NMR is highly challenging. The segmental labelling protocol helps us overcome the problems of spectral overlap with large proteins, but another problem is the reduced sensitivity observed with large proteins due to their reduced tumbling times. This reduction in sensitivity causes amide spectra, which are the standard spectra used in NMR protein experiments, to become too weak to obtain any useful information. In order to overcome this restriction one can turn to methyl based experiments which give much better signal. We obtained a large number of chemical shift assignments for the methyl peaks for each individual domain. These assignments have been transferred to the full length Hsp90 protein and will allow for future experiments.
Significant advances have been made in studying the interactions of Hsp90 with the Glucocorticoid Receptor Ligand Binding Domain(GR-LBD). Previous studies suggested that GR-LBD interacts with Hsp90 only in its apo state and requires Hsp90 for hormone binding. Additionally binding of hormone is supposed to promote dissociation of Hsp90 and GR. However, our studies showed that while there are distinct differences between hormone free and bound GR-LBD, hormone bind GR does indeed interact with Hsp90. We studied these interaction through NMR, combine with Analytical Ultracentrifugation (AUC) with the Buchner group at the Technische Universität München, SAXS with the Madl group at the Technische Universität München and Electron Microscopy (EM) with the Valpuesta group at the Centro Nacional de Biotecnología. We performed these experiments with individual domains, NM tandem domain and full length Hsp90. We showed that GR interacts with a specific binding patch on Hsp90 and consists mainly with the M-domain of Hsp90 and part of the CTD, figure 2. The biochemical experiments performed in collaboration demonstrated that GR-LBD prefers to bind the open conformation of Hsp90 and surprisingly that binding of GR-LBD slows the catalytic cycle of Hsp90. Interestingly our NMR results also demonstrate that hormone bound GR-LBD is present as a structured domain suggesting that Hsp90 can interact with clients which are highly structured but unstable.
These studies have been extended to additional client proteins to obtain a broader idea of the interactions of Hsp90 with client proteins. Similar experiments performed on Tau and Δ131Δ demonstrate different binding regions on Hsp90, figure 2. We have demonstrated that hormone bound GR –LBD forms a structured domain in solution, but both Tau and Δ131Δ are both unstructured. This suggest that Hsp90 interacts with its clients differently depending on their overall structured state. Biochemical results performed by the Buchner group also demonstrate that opposed to GR-LBD binding of Tau increases the catalytic cycle of Hsp90. Suggesting that the behaviour of Hsp90 is highly dependent on the structural characteristics of Hsp90.
Additionally, we have studied small molecule modulators which affect the Hsp90 activity and interactions with its co-chaperones. We have identified a series of compounds which affect the conformational changes of Hsp90 throughout its catalytic cycle. These changes appear to mimic the effects of one of its co-chaperones Aha1. We identified specific regions where these compounds interact with Hsp90 and provide additional information regarding the molecular mechanism of Hsp90 activity and regulation. Another set of compounds binds to the CTD of Hsp90 and promotes the release of GR. These compounds demonstrate a potential treatment for Cushing disease through restoring glucocorticoid sensitivity by targeting Hsp90.
Proper protein folding is essential for cellular homeostasis. Cells employ a large array of chaperones to ensure protein adopt the proper conformation. Hsp90 is an important chaperone and comprises of 1-2% of all protein is eukaryotic cells. Additionally, Hsp90 is upregulated in cancerous cells to facilitate fast proliferation. There are several X-ray crystal structures available for Hsp90 free in solution and interacting with its target proteins. This combined with small angle x-ray/neutron scattering (SAXS/SANS) and Förster resonance energy transfer (FRET) data, demonstrate that HSP90 is a 90 kDa dimeric protein and undergoes large conformational changes throughout its function. However, these crystal structures only provide a static picture of Hsp90 function so the molecular and dynamic details of these rearrangements are not well understood. We employ nuclear magnetic resonance spectroscopy (NMR) as a method to further probe these changes. NMR can provide atomically resolved real time information of Hsp90 in solution either free or interacting with its targets. Hsp90 contains three distinct domains, and we have individually isotopically labeled each domain. This allows each domain to be studied either independently or and in tandem with neighbouring domains. Also a target protein, the glucocorticoid receptor (GR) protein, has been introduced to observe how the different domains are affected by binding to a client protein. Through observing the changes in chemical shifts we have identified specific regions in Hsp90 which are involved in the binding of client proteins. We also employed SAXS/SANS experiments to observe changes in orientation of the domains when interacting with the GR protein. Through combining all of these techniques we have gained a broader understanding of Hsp90 function
We recorded and compared NMR spectra of full length, N-terminal domain (NTD), middle domain (MD), C-terminal domain (CTD) and NTD-MD tandem domains. We optimized a protocol to ligate together two domains using the Sortase Ligation Protein. This procedure allows us to specifically isotopically label individual domains in a multi domain construct and observe it through NMR. The final product contains two domains, one which we can see via NMR and one which is invisible. This greatly reduces spectral overlap so we are better able to analyze the behavior of each domain in a tandem complex, figure 1A. Through comparing the spectral changes between the individual domains and the ligated domain we identified a previously unknown region on NTD which interacts with charged linker region, figure 1B. The overall effects of this interaction on Hsp90 function is still unknown, but studies continue to identify possible effects.
The improved protocol is useful for a wide range of additional biophysical experiments, such as small-angle neutron scattering, electron paramagnetic resonance or fluorescence-based methods. Reducing the number of isotope-labeled amino acids by segmental labeling greatly simplifies NMR spectra. This can allow for the residue-specific analysis of NMR parameters, which may be obscured by signal overlap in a uniformly isotope-labeled sample. The method will be most useful when working with large multi-domain systems, i.e. where flexible linkers connect structurally independent domains.
Studying full length Hsp90 by NMR is highly challenging. The segmental labelling protocol helps us overcome the problems of spectral overlap with large proteins, but another problem is the reduced sensitivity observed with large proteins due to their reduced tumbling times. This reduction in sensitivity causes amide spectra, which are the standard spectra used in NMR protein experiments, to become too weak to obtain any useful information. In order to overcome this restriction one can turn to methyl based experiments which give much better signal. We obtained a large number of chemical shift assignments for the methyl peaks for each individual domain. These assignments have been transferred to the full length Hsp90 protein and will allow for future experiments.
Significant advances have been made in studying the interactions of Hsp90 with the Glucocorticoid Receptor Ligand Binding Domain(GR-LBD). Previous studies suggested that GR-LBD interacts with Hsp90 only in its apo state and requires Hsp90 for hormone binding. Additionally binding of hormone is supposed to promote dissociation of Hsp90 and GR. However, our studies showed that while there are distinct differences between hormone free and bound GR-LBD, hormone bind GR does indeed interact with Hsp90. We studied these interaction through NMR, combine with Analytical Ultracentrifugation (AUC) with the Buchner group at the Technische Universität München, SAXS with the Madl group at the Technische Universität München and Electron Microscopy (EM) with the Valpuesta group at the Centro Nacional de Biotecnología. We performed these experiments with individual domains, NM tandem domain and full length Hsp90. We showed that GR interacts with a specific binding patch on Hsp90 and consists mainly with the M-domain of Hsp90 and part of the CTD, figure 2. The biochemical experiments performed in collaboration demonstrated that GR-LBD prefers to bind the open conformation of Hsp90 and surprisingly that binding of GR-LBD slows the catalytic cycle of Hsp90. Interestingly our NMR results also demonstrate that hormone bound GR-LBD is present as a structured domain suggesting that Hsp90 can interact with clients which are highly structured but unstable.
These studies have been extended to additional client proteins to obtain a broader idea of the interactions of Hsp90 with client proteins. Similar experiments performed on Tau and Δ131Δ demonstrate different binding regions on Hsp90, figure 2. We have demonstrated that hormone bound GR –LBD forms a structured domain in solution, but both Tau and Δ131Δ are both unstructured. This suggest that Hsp90 interacts with its clients differently depending on their overall structured state. Biochemical results performed by the Buchner group also demonstrate that opposed to GR-LBD binding of Tau increases the catalytic cycle of Hsp90. Suggesting that the behaviour of Hsp90 is highly dependent on the structural characteristics of Hsp90.
Additionally, we have studied small molecule modulators which affect the Hsp90 activity and interactions with its co-chaperones. We have identified a series of compounds which affect the conformational changes of Hsp90 throughout its catalytic cycle. These changes appear to mimic the effects of one of its co-chaperones Aha1. We identified specific regions where these compounds interact with Hsp90 and provide additional information regarding the molecular mechanism of Hsp90 activity and regulation. Another set of compounds binds to the CTD of Hsp90 and promotes the release of GR. These compounds demonstrate a potential treatment for Cushing disease through restoring glucocorticoid sensitivity by targeting Hsp90.