Final Report Summary - 14_3_3 (Allosteric effects induced in 14-3-3 targets)
The 14-3-3 family of proteins are associated with oncogenic proteins, such as Raf-1, Bcr-Abl, Bcr, and polyoma middle T-antigen, and are particularly abundant in the mammalian brain where they interact with complexes. These are believed to contribute to a wide range of neurodegenerative disorders, including Creutzfeldt-Jakob, Alzheimer’s, Parkinson’s and polyglutamine repeat diseases. The 14-3-3 proteins function as dimers composed of two ~30-kDa monomers, each capable of binding phosphorylated serine and threonine motifs. Upon binding to the 14-3-3 sites, 14-3-3 induces a conformational change in the target protein allowing for full activity. In this way, 14-3-3s act as a second level of regulation to tightly control the activity of key cellular proteins. Human tyrosine hydroxylase 1 (hTH1) was selected as a typical 14-3-3ζ binding protein for this project. hTH1 is activated by phosphorylation-dependent binding to 14-3-3ζ proteins after phosphorylation of Ser19 and Ser40. The lack of structural information on 14-3-3 in complex with any doubly phosphorylated protein ligand comprising the most common binding modes, precludes a full understanding of 14-3-3 function as well as how activation/deactivation within target proteins is achieved. Therefore, this important problem was addressed by combining advanced computational approaches and NMR spectroscopy in this project.
The proposed computational methodology required incorporation of distance restraints within Hamiltonian replica exchange molecular dynamics (H-REMD). The details about this implementation within the GROMOS05 package and the application of the method for the P450 2D6 complex for which we had a sufficient amount of reference data were published last year. The application of the H-REMD using distance restraints on 14-3-3ζ in complex with double phosphorylated hTH1 led to damaging of C-terminal helixes of 14-3-3ζ and therefore providing unrealistic data. This problem was observed also for another macromolecular system. Our collaborators from Dr. Oostenbrink’s group overcame this problem for aspirin binding to PLA2 by introducing distance-field restraints rather than traditional distance restraints. Our preliminary data about the application of the same approach for the 14-3-3ζ complexes showed big improvements, with the system being kept intact during the H-REMD simulation. In addition to structural information the obtained data allowed us to calculate the free energy profile of the binding of phosphorylated peptides. The progress of H-REMD calculations is quite slow because of the large size of the system and the large number of required replicas. The monitoring of related free energy profiles shows that we have not yet reached converged values, therefore the H-REMD simulations will be continued until the convergence is reached.
Any structural information deduced from NMR on 14-3-3ζ and its complexes require as complete as possible assignments of its 15N,1H HSQC spectrum. It was anticipated that at the time the project began that an assigned 15N,1H HSQC spectrum would be available from our collaborators. Unfortunately, all efforts by our collaborators were unsuccessful. The challenge is two-fold. It is necessary to prepare a large quantity of highly pure, stable, triple labeled 14-3-3ζ protein that will result in a well-resolved NMR spectrum of this 56kDa alpha-helical homodimer. Because backbone assignment of the 15N,1H HSQC spectrum of 14-3-3ζ protein is absolutely essential for the project, it was necessary for Dr. Hritz to take up this challenge. In the first year, a highly pure, stable sample of 14-3-3ζ at a concentration of 1 mM was produced. Because of spectral overlap due to the alpha-helical nature of the protein, it was also necessary to prepare isotopically double (2H, 15N) and triple (2H, 15N, 13C) labeled samples. The purity of the sample was checked by SDS-PAGE and MALDI. The assignment of 85% of the resonances in the 15N, 1H HSQC spectrum of 14-3-3ζ was achieved during the second year of the project.
Another challenging task was preparation of high purity double phosphorylated hTH1_50 peptide without introducing mutants that would alter the natural binding affinity towards 14-3-3ζ. This effort was successful and the optimized procedure containing 12 purification steps allowed us to prepare high purity nonlabeled as well as isotopically labeled [15N, 13C] samples of hTH1 (single and double phosphorylated) in mM concentrations at the beginning of the third year of the project. In order to find the optimal conditions for a study of the complex of 14-3-3ζ with double phosphorylated hTH1 peptide, 31P (1D) NMR titration studies were performed. They revealed a much more complex binding scenario than was so far considered in 14-3-3 field. This data will be submitted for publication very soon. This study also provided us the optimal measuring conditions for high resolution NMR measurements.
Collection of high resolution NMR data of [15N, 13C] hTH1 samples is problematic because of its degradation over the course of a week of NMR measurements. It was therefore very advantageous to apply the non-uniform sampling 5D NMR method that was recently developed in the group of prof. Sklenář and that allows much faster NMR measurements for intrinsically disordered proteins. Currently we are applying this methodology in combination with the preparation of fresh batches of hTH1 for each NMR experiment. We recently assigned backbone and side-chains of hTH1 in free state and are collecting data for the 14-3-3ζ/hTH1 complex with the aim to refine sufficient number of intermolecular NOEs. This data would allow us to refine the structural ensemble of hTH1 obtained by H-REMD described above and provide the most representative structures of such an ensemble.
The procedure of refinement of the structural ensemble through the use of NMR data has much larger applicability. We recently successfully applied this method to the structural ensemble of flexible oligonucleotides in complex with human restriction factor APOBEC3A protein.
One of the aims in the career development plan of Dr. Hritz was to apply his knowledge of computational structural biology in more applied - medicine oriented fields. Therefore a fruitful collaboration was built with Dr. Kate Ryman and Dr. William Klimstra from Center for Vaccine Research at the University of Pittsburgh School of Medicine and with Drs. Vivian Lui and Jennifer Grandis, from the Head and Neck Cancer Center at the Pittsburgh Cancer Institute. Theoretical models constructed through these collaborations explained the effect of mutations modulating the infectivity of the studied alpha viruses and within tyrosine phosphatase, which are abundant in cancer tissues. The results led to three publications (one already published and two more submitted).
In conclusion, there were several severe unexpected problems during this project, particularly the need to assign the 14-3-3ζ protein ourselves that dramatically slowed down the timeline of the project. However, the main line of the project is progressing, and we believe that the last and main aim of this project, the determination of representative structures of the 14-3-3ζ/hTH1 complex, will be finalized next year.
The proposed computational methodology required incorporation of distance restraints within Hamiltonian replica exchange molecular dynamics (H-REMD). The details about this implementation within the GROMOS05 package and the application of the method for the P450 2D6 complex for which we had a sufficient amount of reference data were published last year. The application of the H-REMD using distance restraints on 14-3-3ζ in complex with double phosphorylated hTH1 led to damaging of C-terminal helixes of 14-3-3ζ and therefore providing unrealistic data. This problem was observed also for another macromolecular system. Our collaborators from Dr. Oostenbrink’s group overcame this problem for aspirin binding to PLA2 by introducing distance-field restraints rather than traditional distance restraints. Our preliminary data about the application of the same approach for the 14-3-3ζ complexes showed big improvements, with the system being kept intact during the H-REMD simulation. In addition to structural information the obtained data allowed us to calculate the free energy profile of the binding of phosphorylated peptides. The progress of H-REMD calculations is quite slow because of the large size of the system and the large number of required replicas. The monitoring of related free energy profiles shows that we have not yet reached converged values, therefore the H-REMD simulations will be continued until the convergence is reached.
Any structural information deduced from NMR on 14-3-3ζ and its complexes require as complete as possible assignments of its 15N,1H HSQC spectrum. It was anticipated that at the time the project began that an assigned 15N,1H HSQC spectrum would be available from our collaborators. Unfortunately, all efforts by our collaborators were unsuccessful. The challenge is two-fold. It is necessary to prepare a large quantity of highly pure, stable, triple labeled 14-3-3ζ protein that will result in a well-resolved NMR spectrum of this 56kDa alpha-helical homodimer. Because backbone assignment of the 15N,1H HSQC spectrum of 14-3-3ζ protein is absolutely essential for the project, it was necessary for Dr. Hritz to take up this challenge. In the first year, a highly pure, stable sample of 14-3-3ζ at a concentration of 1 mM was produced. Because of spectral overlap due to the alpha-helical nature of the protein, it was also necessary to prepare isotopically double (2H, 15N) and triple (2H, 15N, 13C) labeled samples. The purity of the sample was checked by SDS-PAGE and MALDI. The assignment of 85% of the resonances in the 15N, 1H HSQC spectrum of 14-3-3ζ was achieved during the second year of the project.
Another challenging task was preparation of high purity double phosphorylated hTH1_50 peptide without introducing mutants that would alter the natural binding affinity towards 14-3-3ζ. This effort was successful and the optimized procedure containing 12 purification steps allowed us to prepare high purity nonlabeled as well as isotopically labeled [15N, 13C] samples of hTH1 (single and double phosphorylated) in mM concentrations at the beginning of the third year of the project. In order to find the optimal conditions for a study of the complex of 14-3-3ζ with double phosphorylated hTH1 peptide, 31P (1D) NMR titration studies were performed. They revealed a much more complex binding scenario than was so far considered in 14-3-3 field. This data will be submitted for publication very soon. This study also provided us the optimal measuring conditions for high resolution NMR measurements.
Collection of high resolution NMR data of [15N, 13C] hTH1 samples is problematic because of its degradation over the course of a week of NMR measurements. It was therefore very advantageous to apply the non-uniform sampling 5D NMR method that was recently developed in the group of prof. Sklenář and that allows much faster NMR measurements for intrinsically disordered proteins. Currently we are applying this methodology in combination with the preparation of fresh batches of hTH1 for each NMR experiment. We recently assigned backbone and side-chains of hTH1 in free state and are collecting data for the 14-3-3ζ/hTH1 complex with the aim to refine sufficient number of intermolecular NOEs. This data would allow us to refine the structural ensemble of hTH1 obtained by H-REMD described above and provide the most representative structures of such an ensemble.
The procedure of refinement of the structural ensemble through the use of NMR data has much larger applicability. We recently successfully applied this method to the structural ensemble of flexible oligonucleotides in complex with human restriction factor APOBEC3A protein.
One of the aims in the career development plan of Dr. Hritz was to apply his knowledge of computational structural biology in more applied - medicine oriented fields. Therefore a fruitful collaboration was built with Dr. Kate Ryman and Dr. William Klimstra from Center for Vaccine Research at the University of Pittsburgh School of Medicine and with Drs. Vivian Lui and Jennifer Grandis, from the Head and Neck Cancer Center at the Pittsburgh Cancer Institute. Theoretical models constructed through these collaborations explained the effect of mutations modulating the infectivity of the studied alpha viruses and within tyrosine phosphatase, which are abundant in cancer tissues. The results led to three publications (one already published and two more submitted).
In conclusion, there were several severe unexpected problems during this project, particularly the need to assign the 14-3-3ζ protein ourselves that dramatically slowed down the timeline of the project. However, the main line of the project is progressing, and we believe that the last and main aim of this project, the determination of representative structures of the 14-3-3ζ/hTH1 complex, will be finalized next year.