Final Activity Report Summary - FOLDING AND MISFOLDING (Development of time resolved resonance energy transfer (tr-FRET) based methods for investigation of ... protein folding and misfolding)
Scientific background proteins are the machines of life. The main content of the genome of every organism is a set of instruction for the construction of the structure of protein molecules. This capacity if the main precondition for the reproduction of any organism on earth. Proteins are linear chain molecules made of amino acids which under physiological conditions fold into specific structure which can perform life supporting functions. Eighty-five years ago Anson and Mirsky showed that proteins can refold spontaneously. It was subsequently shown by Christian Anfinsen in the late fifties of the 20th century that all the information needed for this complex transition is contained in the amino acids sequence of each protein1.
Nevertheless, the mechanism of this surprisingly fast, efficient and accurate chain folding process, 'the protein folding problem', remains unresolved. The long range goal of the research into the problem is to be "able to read genes", i.e. to be able to predict native structure of a protein based on the genetic information and the thermodynamic conditions of the system. In principle, we should be able to produce an algorithm that mimics the search for the most stable conformation of a polypeptide chain under specified conditions.
Yet, despite impressive progress, this predictive ability is still not solid. This is probably due to the interdependence of multiple interatomic interactions that cannot be accounted for by current energy calculations Apart from difficulties of solving such a complex many body problem, and the correlations between huge number of interactions that limit the ab-initio computational prediction, it seems that the genetic information also encodes a mechanism for construction of the native structure. The paradox of the complexity of the folding transition versus the fast rate of its completion led to a search for a mechanism that reduces the stochastic nature of the search for local or global energy minima. In the cytoplasm a slowly folding protein molecules might either be captured by the "recycling" mechanism of protein degradation or misfold rather than fold. Therefore the mechanism that accelerates the rate of folding is a mode of control of the fate of protein molecules.
Complex molecular machinery in the cell is available to avoid misfolding and further accelerate the folding of proteins in vivo. Local and non-local interactions in the early folding steps even for short polypeptide chains, the number of possible conformations is astronomically large, and therefore an undirected search for the native conformation would be very slow. Rather, processes must occur early in folding which exclude access to most of the (unproductive) conformational space and rapidly narrow the folding reaction to a small number of potentially productive routes. Several general mechanisms of protein folding have been proposed which relate to the initial reduction of the dimensions of the chain by solvent exclusion followed by stepwise or concerted formation of sub-domain elements of the chain.
A major challenge for both the design of experiments and modelling detailed mechanisms of folding concerns the order and relative importance of chain collapse and (secondary) structure formation during the initial stages of folding. Recent advances in the ability to monitor folding indicate that collapse and formation of native contacts can occur simultaneously. Detailed monitoring of the folding transition at the atomic level or even at the sub-domain level might reveal hidden intermediates that common probes do not resolve. Given this basic framework for protein folding, we can begin to identify the limiting factors that affect protein folding. A number of investigations focused on rates of formation of initial contacts in disordered polypeptides in an effort to determine the upper limit for protein folding rates. Formation of non-local contacts changes the local topology permitting more rapid formation of subsequent contacts. Similar conclusions were also derived from our studies of the folding of five model proteins by means of time resolved dynamic non-radiative excitation energy transfer (tr-FRET).
Ensembles of the denatured and partially folded molecules were defined in terms of the mean and width of selected (by labelling) intramolecular distances and the rates of their fast fluctuations. Detection of sub-populations of partially folded molecules with native like non-local contacts by means of time resolved FRET experiments led us to the "loop hypothesis"; we suggest that the formation of a few, specific non-local contacts that effectively crosslink the still solvated polypeptide chain may be one of the basic "tricks" used by protein molecules in order to overcome the well-known "Levinthal paradox".
We hypothesise that non-local interactions are effective in the early steps of the folding transition. Thus, it seems that there must be a delicate balance between local and non-local interactions, which might be different for different proteins and different protein types. The challenge of detection and characterisation of the early formed sub domain structures It is known that the transition state ensemble (TSE) of globular proteins is native like. Thus, the major steps of overcoming the configurational entropy of the backbone occurs prior to the transition state. The earliest events in protein folding are thought to play a key role in directing the folding process, but are experimentally inaccessible, because they occur so rapidly and involve large conformational ensembles with widely variable structures and poorly populated intermediate structures.
We have shown that the combination of site specific fluorescence labelling and time resolved FRET measurements (in the equilibrium and kinetics modes) is unique in that distributions of sub-domain intramolecular distances can be determined at a time resolution that is limited only by the fast initiation (pressure, temperature jump, or mixing) methods. We have further shown that nanosecond fluctuations of segmental end to end in disordered proteins can be detected, and subpopulations of conformers can be resolved. The central objective of the present project was the development of new methods and instruments for monitoring the fast formation of non-local contacts within the initial phases of the refolding transition of AK.
We developed methods and experimental approaches for structural characterisation of equilibrium and transient ensembles of model protein molecules during the early phases of their folding transitions. We combined kinetic and equilibrium biophysical methods for characterisation of distributions of global and sub-domain structural parameters in an effort aimed at identifying the key components of the initial folding steps. These new methods when combined with perturbation mutations enable us now to focus on identifying the interacting residues with the long term goal of characterising the types of interactions and types of sequence elements that contribute them.
Based on the achievements of the present projects we are now continuing in an effort to discover specific relationship between sequence-interactions and sub-domain structures. Once these are identified, the folding process can be modelled as a directed process based on elementary stochastic steps. In the present project we focused at two main objectives: first the application of biophysical (mainly spectroscopic) approaches, combined with rapidly switched structure perturbation techniques and protein engineering methods for characterisation of the global and segmental folds of the backbone of model proteins during the earliest phases of the folding transition. The second main objective of the research program was to characterise the initial conformational changes that lead to amyloid aggregation of natively unfolded proteins. We used the above novel technologies in a study of the solution conformation of a so called "natively unfolded protein" in order to reveal the conformational changes that precede the aggregation.
We believe that by understanding the factors that affect the initiation of aggregation we might also suggest means for stabilising the soluble forms of the amyloid forming proteins and thus slow down the process and help those who suffer the devastating neurodegenerative diseases. Therefore we studied the basic conformational characteristics of a model protein under folding conditions. In the framework of these two objectives we performed the following specific projects:
1. As part of the first objective we have developed new approach for monitoring subpopulations of conformers of globular proteins in solution. We are using time resolved FRET measurements which are sensitive enough to resolve and characterise the distributions of intramolecular distances of two subpopulations of conformers of partially folded protein. This method enabled us to show that the folding of small globular protein is a barrier crossing mechanism and not a downhill process.
2. We then turned our efforts to the development of three ultrafast FRET detected kinetic experiments for monitoring the earliest phases of the folding of globular proteins.
The following new instruments and method were developed:
(a) New stopped flow double kinetics method with picosecond time resolution of the spectroscopic time regime was developed. The instrument is based on a fast mixer; dual wavelength detection; picosecond laser excitation source and ultrafast (13GHz) fast digitizer oscilloscope. Using this instrument we have discovered that at least one long 44 residue long loop in the molecule of the protein adenylate kinase (AK) is formed within one millisecond from the initiation of refolding of the protein. Additional loops are now being tested. We next turned to resolve the time scale of closure of the early loops detected by the stopped flow based experiments. There are two main ways to induce very fast initiation of the refolding of globular proteins in solution. These are perturbation of the equilibrium conditions of an ensemble of protein molecules by changes of either the temperature or the hydrostatic pressure. We developed two new instruments based on these perturbation approaches.
(b) First we built a new 500 bar pressure jump machine for FRET detection of unfolding/refolding transitions of proteins. The machine has microsecond time resolution. The heart of the machine is a very sophisticated pressure cell which was produced by the Bar Ilan University machine shop. The instrument was tested and is now being used for studies of the loop closure mechanism in the folding of the AK molecule.
(c) In parallel we have built a new temperature jump machine for double kinetics unfolding/refolding experiments. The machine is based on a unique powerful IR laser for initiation of the unfolding or refolding of the protein by 10 to 20 oC change of temperature and a probe laser for time resolved FRET detection. The machine was tested and is now being used for folding studies.
(d) Another line of development included the development of FRET detected single molecule spectroscopic methods for studying folding and misfolding of globular proteins. The machine was installed, calibrated and tested. New double labeled mutants were prepared and measured. Ultrafast fluctuations of the distance between the LID and the CORE domains of the AK molecule were detected by FRET FCS measurements.
(e) Integration of the equilibrium and kinetics tr-FRET data (items a-e above) with mutational perturbations data (F analysis) is now being done in order identify putative key residues or clusters of residues assumed to form key intramolecular contacts in the ensembles of partially folded molecules. Folding perturbation mutations were introduced in the labeled AK mutants at one or two of these putative key residue in search for the steps in the folding process which are affected and in order to map the key sequence elements responsible for the fast initial refolding steps. One such cluster was identified and mutants there in are now being tested by the fast kinetics methods.
(f) The second overall objective of the present study was aimed at studying the conformational ensembles of an intrinsically unfolded protein. We used the molecule of human ?-synuclein as the model protein. We developed methods for producing double labeled mutants of this protein and keeping them in the monomeric state for the duration of the measurements. The disordered state of natural polypeptides is the starting point for all refolding experiments and therefore characterisation of the ensembles of globular protein molecules in their fully or partially disordered states is essential for understanding the mechanism of folding. Understanding either the mechanism of action or of misfolding, i.e. aggregation of the large group of proteins known to be intrinsically disordered under native conditions also depends on structural characterisation of the ensembles of such protein molecules. ?-Synuclein (?S), an abundant 140-residue neuronal protein of unknown function, is a typical intrinsically disordered protein. It is the primary component of the fibrillar inclusions found in the brain of Parkinson disease patients. We have applied time resolved fluorescence resonance energy transfer experiments and determined the end-to-end distance distributions of eight labeled chain segments in the N terminal and NAC domains in the ?-synuclein molecules. The mean and full width at half maximum of the intramolecular distance distributions of the labeled chain segments were temperature independent over the temperature range from 5 to 40 oC and their ratio was unity. These are characteristics of ensembles of disordered polypeptide molecules. The intramolecular diffusion coefficient of the labeled segment ends relative to each other which were determined by analysis of the trFRET experiments. Very fast nanosecond dynamics of the internal chain segments was obtained. Its variation between zero and 30 ?2/ns revealed subtle conformational constraints in some of the segments. These experiments yield a unique data set that enabled us to examine the chain length dependence of the end to end distances of a disordered natural protein under folding conditions and compare with theoretical models. Overall the ?S molecule shows characteristics predicted for disordered polypeptide with a bias towards higher dimensions due to the excluded volume effect or mild repulsive interactions. The most external N terminal chain sections and a section in the NAC domain showed small deviation from disordered ensemble characteristics and damped rates of intramolecular diffusion. These might be foci of initiation of folding into aggregation prone conformations and sites that determine the limited stability of the monomeric form.
(g) A computational study which complements the experimental work tested the extent of conservation of local and non-local interactions in globular proteins whose structures were determined and presented to the protein data bank. The results of this study support our working hypothesis that non local interactions are dominant in the folding transition since it was found that non local interactions are more conserved than local interactions.
Nevertheless, the mechanism of this surprisingly fast, efficient and accurate chain folding process, 'the protein folding problem', remains unresolved. The long range goal of the research into the problem is to be "able to read genes", i.e. to be able to predict native structure of a protein based on the genetic information and the thermodynamic conditions of the system. In principle, we should be able to produce an algorithm that mimics the search for the most stable conformation of a polypeptide chain under specified conditions.
Yet, despite impressive progress, this predictive ability is still not solid. This is probably due to the interdependence of multiple interatomic interactions that cannot be accounted for by current energy calculations Apart from difficulties of solving such a complex many body problem, and the correlations between huge number of interactions that limit the ab-initio computational prediction, it seems that the genetic information also encodes a mechanism for construction of the native structure. The paradox of the complexity of the folding transition versus the fast rate of its completion led to a search for a mechanism that reduces the stochastic nature of the search for local or global energy minima. In the cytoplasm a slowly folding protein molecules might either be captured by the "recycling" mechanism of protein degradation or misfold rather than fold. Therefore the mechanism that accelerates the rate of folding is a mode of control of the fate of protein molecules.
Complex molecular machinery in the cell is available to avoid misfolding and further accelerate the folding of proteins in vivo. Local and non-local interactions in the early folding steps even for short polypeptide chains, the number of possible conformations is astronomically large, and therefore an undirected search for the native conformation would be very slow. Rather, processes must occur early in folding which exclude access to most of the (unproductive) conformational space and rapidly narrow the folding reaction to a small number of potentially productive routes. Several general mechanisms of protein folding have been proposed which relate to the initial reduction of the dimensions of the chain by solvent exclusion followed by stepwise or concerted formation of sub-domain elements of the chain.
A major challenge for both the design of experiments and modelling detailed mechanisms of folding concerns the order and relative importance of chain collapse and (secondary) structure formation during the initial stages of folding. Recent advances in the ability to monitor folding indicate that collapse and formation of native contacts can occur simultaneously. Detailed monitoring of the folding transition at the atomic level or even at the sub-domain level might reveal hidden intermediates that common probes do not resolve. Given this basic framework for protein folding, we can begin to identify the limiting factors that affect protein folding. A number of investigations focused on rates of formation of initial contacts in disordered polypeptides in an effort to determine the upper limit for protein folding rates. Formation of non-local contacts changes the local topology permitting more rapid formation of subsequent contacts. Similar conclusions were also derived from our studies of the folding of five model proteins by means of time resolved dynamic non-radiative excitation energy transfer (tr-FRET).
Ensembles of the denatured and partially folded molecules were defined in terms of the mean and width of selected (by labelling) intramolecular distances and the rates of their fast fluctuations. Detection of sub-populations of partially folded molecules with native like non-local contacts by means of time resolved FRET experiments led us to the "loop hypothesis"; we suggest that the formation of a few, specific non-local contacts that effectively crosslink the still solvated polypeptide chain may be one of the basic "tricks" used by protein molecules in order to overcome the well-known "Levinthal paradox".
We hypothesise that non-local interactions are effective in the early steps of the folding transition. Thus, it seems that there must be a delicate balance between local and non-local interactions, which might be different for different proteins and different protein types. The challenge of detection and characterisation of the early formed sub domain structures It is known that the transition state ensemble (TSE) of globular proteins is native like. Thus, the major steps of overcoming the configurational entropy of the backbone occurs prior to the transition state. The earliest events in protein folding are thought to play a key role in directing the folding process, but are experimentally inaccessible, because they occur so rapidly and involve large conformational ensembles with widely variable structures and poorly populated intermediate structures.
We have shown that the combination of site specific fluorescence labelling and time resolved FRET measurements (in the equilibrium and kinetics modes) is unique in that distributions of sub-domain intramolecular distances can be determined at a time resolution that is limited only by the fast initiation (pressure, temperature jump, or mixing) methods. We have further shown that nanosecond fluctuations of segmental end to end in disordered proteins can be detected, and subpopulations of conformers can be resolved. The central objective of the present project was the development of new methods and instruments for monitoring the fast formation of non-local contacts within the initial phases of the refolding transition of AK.
We developed methods and experimental approaches for structural characterisation of equilibrium and transient ensembles of model protein molecules during the early phases of their folding transitions. We combined kinetic and equilibrium biophysical methods for characterisation of distributions of global and sub-domain structural parameters in an effort aimed at identifying the key components of the initial folding steps. These new methods when combined with perturbation mutations enable us now to focus on identifying the interacting residues with the long term goal of characterising the types of interactions and types of sequence elements that contribute them.
Based on the achievements of the present projects we are now continuing in an effort to discover specific relationship between sequence-interactions and sub-domain structures. Once these are identified, the folding process can be modelled as a directed process based on elementary stochastic steps. In the present project we focused at two main objectives: first the application of biophysical (mainly spectroscopic) approaches, combined with rapidly switched structure perturbation techniques and protein engineering methods for characterisation of the global and segmental folds of the backbone of model proteins during the earliest phases of the folding transition. The second main objective of the research program was to characterise the initial conformational changes that lead to amyloid aggregation of natively unfolded proteins. We used the above novel technologies in a study of the solution conformation of a so called "natively unfolded protein" in order to reveal the conformational changes that precede the aggregation.
We believe that by understanding the factors that affect the initiation of aggregation we might also suggest means for stabilising the soluble forms of the amyloid forming proteins and thus slow down the process and help those who suffer the devastating neurodegenerative diseases. Therefore we studied the basic conformational characteristics of a model protein under folding conditions. In the framework of these two objectives we performed the following specific projects:
1. As part of the first objective we have developed new approach for monitoring subpopulations of conformers of globular proteins in solution. We are using time resolved FRET measurements which are sensitive enough to resolve and characterise the distributions of intramolecular distances of two subpopulations of conformers of partially folded protein. This method enabled us to show that the folding of small globular protein is a barrier crossing mechanism and not a downhill process.
2. We then turned our efforts to the development of three ultrafast FRET detected kinetic experiments for monitoring the earliest phases of the folding of globular proteins.
The following new instruments and method were developed:
(a) New stopped flow double kinetics method with picosecond time resolution of the spectroscopic time regime was developed. The instrument is based on a fast mixer; dual wavelength detection; picosecond laser excitation source and ultrafast (13GHz) fast digitizer oscilloscope. Using this instrument we have discovered that at least one long 44 residue long loop in the molecule of the protein adenylate kinase (AK) is formed within one millisecond from the initiation of refolding of the protein. Additional loops are now being tested. We next turned to resolve the time scale of closure of the early loops detected by the stopped flow based experiments. There are two main ways to induce very fast initiation of the refolding of globular proteins in solution. These are perturbation of the equilibrium conditions of an ensemble of protein molecules by changes of either the temperature or the hydrostatic pressure. We developed two new instruments based on these perturbation approaches.
(b) First we built a new 500 bar pressure jump machine for FRET detection of unfolding/refolding transitions of proteins. The machine has microsecond time resolution. The heart of the machine is a very sophisticated pressure cell which was produced by the Bar Ilan University machine shop. The instrument was tested and is now being used for studies of the loop closure mechanism in the folding of the AK molecule.
(c) In parallel we have built a new temperature jump machine for double kinetics unfolding/refolding experiments. The machine is based on a unique powerful IR laser for initiation of the unfolding or refolding of the protein by 10 to 20 oC change of temperature and a probe laser for time resolved FRET detection. The machine was tested and is now being used for folding studies.
(d) Another line of development included the development of FRET detected single molecule spectroscopic methods for studying folding and misfolding of globular proteins. The machine was installed, calibrated and tested. New double labeled mutants were prepared and measured. Ultrafast fluctuations of the distance between the LID and the CORE domains of the AK molecule were detected by FRET FCS measurements.
(e) Integration of the equilibrium and kinetics tr-FRET data (items a-e above) with mutational perturbations data (F analysis) is now being done in order identify putative key residues or clusters of residues assumed to form key intramolecular contacts in the ensembles of partially folded molecules. Folding perturbation mutations were introduced in the labeled AK mutants at one or two of these putative key residue in search for the steps in the folding process which are affected and in order to map the key sequence elements responsible for the fast initial refolding steps. One such cluster was identified and mutants there in are now being tested by the fast kinetics methods.
(f) The second overall objective of the present study was aimed at studying the conformational ensembles of an intrinsically unfolded protein. We used the molecule of human ?-synuclein as the model protein. We developed methods for producing double labeled mutants of this protein and keeping them in the monomeric state for the duration of the measurements. The disordered state of natural polypeptides is the starting point for all refolding experiments and therefore characterisation of the ensembles of globular protein molecules in their fully or partially disordered states is essential for understanding the mechanism of folding. Understanding either the mechanism of action or of misfolding, i.e. aggregation of the large group of proteins known to be intrinsically disordered under native conditions also depends on structural characterisation of the ensembles of such protein molecules. ?-Synuclein (?S), an abundant 140-residue neuronal protein of unknown function, is a typical intrinsically disordered protein. It is the primary component of the fibrillar inclusions found in the brain of Parkinson disease patients. We have applied time resolved fluorescence resonance energy transfer experiments and determined the end-to-end distance distributions of eight labeled chain segments in the N terminal and NAC domains in the ?-synuclein molecules. The mean and full width at half maximum of the intramolecular distance distributions of the labeled chain segments were temperature independent over the temperature range from 5 to 40 oC and their ratio was unity. These are characteristics of ensembles of disordered polypeptide molecules. The intramolecular diffusion coefficient of the labeled segment ends relative to each other which were determined by analysis of the trFRET experiments. Very fast nanosecond dynamics of the internal chain segments was obtained. Its variation between zero and 30 ?2/ns revealed subtle conformational constraints in some of the segments. These experiments yield a unique data set that enabled us to examine the chain length dependence of the end to end distances of a disordered natural protein under folding conditions and compare with theoretical models. Overall the ?S molecule shows characteristics predicted for disordered polypeptide with a bias towards higher dimensions due to the excluded volume effect or mild repulsive interactions. The most external N terminal chain sections and a section in the NAC domain showed small deviation from disordered ensemble characteristics and damped rates of intramolecular diffusion. These might be foci of initiation of folding into aggregation prone conformations and sites that determine the limited stability of the monomeric form.
(g) A computational study which complements the experimental work tested the extent of conservation of local and non-local interactions in globular proteins whose structures were determined and presented to the protein data bank. The results of this study support our working hypothesis that non local interactions are dominant in the folding transition since it was found that non local interactions are more conserved than local interactions.