The Folding Energy Landscape of The Folding Energy Landscape of The Folding Energy Landscape of The Folding Energy Landscape of

MerPMerPMerPMerP

AnnAnnAnnAnn----Christin BrorssonChristin BrorssonChristin BrorssonChristin Brorsson

______________________________________________ Department of BiochemistryDepartment of BiochemistryDepartment of BiochemistryDepartment of Biochemistry

Umeå UniversityUmeå UniversityUmeå UniversityUmeå University

Umeå, SwedenUmeå, SwedenUmeå, SwedenUmeå, Sweden

2004200420042004

Copyright by Ann-Christin Brorsson Front cover: The 3-dimensional structure of MerP Printed in Sweden by Solfjädern Offset AB, Umeå 2004 ISBN 91-7305-710-X

The Folding Energy Landscape of The Folding Energy Landscape of The Folding Energy Landscape of The Folding Energy Landscape of

MerPMerPMerPMerP

ByByByBy

AnnAnnAnnAnn----Christin BrorssonChristin BrorssonChristin BrorssonChristin Brorsson

Department of BiochemistryDepartment of BiochemistryDepartment of BiochemistryDepartment of Biochemistry

Umeå UniversityUmeå UniversityUmeå UniversityUmeå University

Umeå, SwedenUmeå, SwedenUmeå, SwedenUmeå, Sweden

2004200420042004

Akademisk avhandling

Som med vederbörligt tillstånd av Rektorsämbetet vid Umeå universitet för

avläggande av filosofie doktorsexamen vid teknisk-naturvetenskapliga

fakulteten kommer att offentligen försvaras på Kemiska institutionen, Umeå

universitet, hörsal KB3A9, KBC, fredagen den 1 oktober 2004, kl. 10.00.

Fakultetsopponent: Prof. Daniel Otzen

Institute of Life Sciences

Section of Biotechnology

Aalborg University

Denmark

Abstract This thesis is based on studies, described in four papers, in which the folding energy landscape of MerP was investigated by various techniques. MerP is a water-soluble 72 amino acid protein with a secondary structure consisting of four anti-parallel β-strands and two α-helices on one side of the sheet in the order β1α1β2β3α2β4. The first paper describes the use of CD and fluorescence analysis to examine the folding/unfolding process of MerP. From these experiments it was found that the protein folds according to a two-state model in which only the native and unfolded forms are populated without any visible intermediates. With a rate constant of 1.2 s-1, the folding rate was found to be unusually slow for a protein of this size. The studies presented in the second and third papers were based on measurements of native-state amide proton exchange at different temperatures (Paper II) and GuHCl concentrations (Paper III) in the pre-transitional region. In these studies partially unfolded forms were found for MerP which are essentially unrelated to each other. Thus, in the folding energy landscape of MerP, several intermediates seem to occur on different folding trajectories that are parallel to each other. The slow folding rate of MerP might be coupled to extensive visitation of these conformations. Hydrogen exchange in MerP did also reveal structure-dependent differences in compactness between the denatured states in GuHCl and H2O. In the last paper multivariate data analysis was applied to 2-dimensional NMR data to detect conformational changes in the structure of MerP induced by GuHCl. From this analysis it was suggested that regions involved in the most flexible part of the protein structure are disrupted at rather low denaturant concentrations (< 2.1 M GuHCl) while the native structures of the most stable parts are still not completely ruptured at 2.9 M GuHCl. Finally, the stability, kinetics, contact order and folding nuclei of six proteins with similar topology (MerP, U1A, S6, ADA2h, AcP and HPr) were compared. In this analysis it was found that their folding properties are quite diverse, despite their topological similarities, and no general rules that have been formulated yet can adequately predict their folding behaviour. Keywords: protein folding and stability; hydrogen exchange; intermediate; partial unfolding

ISBN 91-7305-710-X

Papers in this thesis This thesis is based on the following papers which are included as supplements after the thesis. In the text, they will be referred to by their roman numerals. I. Aronsson, G., Brorsson, A-C., Sahlman, L. and Jonsson, B-H.

(1997) Remarkably slow folding of a small protein, MerP. FEBS Lett. 411, 359-364.

II. Brorsson, A-C., Kjellson, A., Aronsson, G., Sethson, I., Hambraeus,

C. and Jonsson, B-H. (2004) The “Two-State folder” MerP forms partially unfolded structures that show temperature dependent hydrogen exchange. J. Mol. Biol. 340, 333-344.

III. Brorsson, A-C., Sethson, I. and Jonsson, B-H. (2004) Hydrogen

exchange in MerP reveals structure-dependent differences in compactness between the denatured states in GuHCl and H2O. Submitted to Journal of Molecular Biology.

IV. Berglund, A., Brorsson, A-C., Jonsson, B-H. and Sethson, I. (2004)

The equilibrium unfolding of MerP characterized by multivariate analysis of 2D NMR data. Submitted to Journal of Magnetic Resonance.

Abbreviations MerP Mercury binding Protein CD Circular Dichroism GuHCl Guanidine HydroChloride MVDA Multi Variate Data Analysis NMR Nuclear Magnetic Resonance PCA Principal Component Analysis PLS Partial Least Squares U1A RNA binding domain of the U1 small nuclear

ribonucleoprotein A S6 Ribosomal protein S6 ADA2h Activation Domain of procarboxypeptidase A2 AcP Muscle AcylPhosphatase HPr Histidine-containing Phosphocarrier protein CspB Cold shock protein B DNA DeoxyriboNucleinAcid UV UltraViolet

Table of Contents Introduction 1 Historical background of proteins 1 Protein structure and function 1 The protein folding problem 4 Conformational properties of proteins 5 Protein stability 5 Protein folding 7 Protein unfolding 14 Protein misfolding 17 Techniques used in this study 18 Nuclear Magnetic Resonance 18 Hydrogen Exchange 21 Fluorescence Spectroscopy 27 Circular Dichroism Spectroscopy 27 Multivariate Data Analysis 28 The protein MerP 29 Summary of papers 30 I Fluorescence and CD analysis of the folding properties of MerP 30 II Native-state HX of MerP at increasing temperatures 33 III Native-state HX of MerP at increasing concentrations of GuHCl 38 IV Multivariate analysis of 2D NMR data 42 Comparison of proteins with similar topology to MerP 44 Conclusions 46 Acknowledgements 47 References 48 Papers I-IV

1

Introduction Historical background of proteins Four classes of molecules constitute the main components in living systems: proteins, carbohydrates, lipids, and nucleic acids. Of these components proteins have the widest diversity in structure and function. Proteins were first discovered in 1838 by the Swedish scientist Jöns Berzelius. The original meaning of proteins is first thing, derived from the Greek word for “pre-eminent”. Over the following century it was found that the building blocks in proteins are made of 20 different amino acids linked together in an unbranched biopolymer known as a polypeptide chain. In the early 1950s, Linus Pauling and Robert Corey proposed two possible hydrogen-bonded configurations for the polypeptide chain: a helical formation,1 and a pleated-sheet configuration2 today known as an α-helix and a β-strand, respectively. At almost the same time Fred Sanger, a Cambridge scientist, managed to develop new methods for amino acid sequencing and was able to determine the complete sequence of insulin (a hormone protein).3 The first detailed protein structures were solved, in 1957, of myoglobin (a protein that stores oxygen) by John Kendrew and colleagues4 and in 1959 of haemoglobin (an oxygen transport protein) by Perutz and his collaborators.5 Protein structure and function An amino acid contains both amino and carboxylic functional groups. In the polypeptide chain the amino acids (also called residues) are linked together by a peptide bond, and one molecule of water is released with the formation of each peptide link. Figure 1 shows a schematic diagram of two amino acids linked together by a peptide bond. The only part that differs between the various amino acids is their side chains (labelled R1 and R2 for the respective amino acids in Figure 1) except for proline, which has a side chain covalently linked to the amide nitrogen and therefore lacks an amino group. According to the chemical properties of the side chain the amino acids are classified into four different groups: acidic, basic, hydrophilic (polar) and hydrophobic (non-polar).

2

Figure 1. Schematic drawing of two amino acids linked together by a peptide bond. Protons are coloured white, carbon atoms grey, oxygen atoms black and nitrogen atoms light grey. The side chains of the amino acid are labelled R1 and R2 for the amino acids to the left and right, respectively. Before a protein can function it has to fold into a unique 3-dimensional structure called its native state. The protein structure can be described at four levels:

• The primary structure is the order of the amino acids in the polypeptide chain.

• The secondary structure is composed of repetitive 3-dimensional conformational units. The key to this structure is hydrogen bonding between the amide protons and carbonyl groups of specific pairs of residues. Most frequently the secondary structure constitutes either an α-helix or a β-sheet structure. In an α-helix the structure is stabilized by a series of hydrogen bonds from C=O of residue n to N-H of residue n + 4 (see Figure 2a), while in the β-sheet the structure is stabilized by hydrogen bonds between the C=O and N-H groups of two adjacent β-strands (see Figure 2b). Very often the structural units between elements of secondary structure constitute well-defined (non-repeating) turn structures.

Amino group

Peptide group Peptide bond

Carboxyl group

Amide proton

3

a b

Figure 2. Schematic drawings of secondary structural units. (a) α-helix, (b) β-sheet. The colour scheme is the same as described in Figure 1. Hydrogen bonds are shown by dotted lines. For clarity only the backbone atoms are shown.

• The tertiary structure is the overall 3-dimensional configuration of

the polypeptide chain, which is stabilised by interactions between the side chains of the residues. The tertiary structure can be divided into one or more subdomains that constitute local parts with specific tertiary structure elements.

• The quaternary structure is the overall configuration adopted by proteins consisting of more than one polypeptide, which is stabilised by interactions between the polypeptides.

The 3-dimensional structure of a protein can be determined by X-ray diffraction or Nuclear Magnetic Resonance. Figure 3 shows the 3-dimesional structure of MerP.

Figure 3. The 3-dimensional structure of MerP.

4

The diversity of the biological functions of proteins is enormous, covering a vast range of properties. For example, there are: transport proteins like haemoglobin (which carries oxygen to our tissues), hormones like insulin (which signals our bodies to store excess sugar), storage proteins like myoglobin (which stores oxygen in the muscles), enzymes (which catalyse chemical reactions), contractile proteins like actin and myosin (involved in muscle contraction), receptor proteins that allow signals to pass through membranes, structural proteins like the collagen in our tendons and ligaments, and proteins involved in our immune defence system (antibodies) etc. In the human body there are between 30 000 and 40 000 genes that encode the different proteins responsible for a vast array of biological activities. The protein folding problem Today we have a thorough understanding of how cells construct sequences of amino acids using the DNA as a template. There are also methods to analyze the amino acid contents in specific proteins, and to determine the order of the residues. However, there are still major uncertainties about how polypeptide chains are able to fold into the well defined structures necessary for proteins to perform their specific functions. In the late 1950s Christian Anfinsen showed that the amino acid sequence contains enough information for the polypeptide to fold into a specific three-dimensional structure and today it is known that for many proteins all the information needed for them to correctly fold into their biologically active shapes lies in the primary amino acid sequence.6 But the question remains, how is it possible for an unfolded protein, with hundreds of millions of potential folded states to choose from, to find the correct fold in only a few seconds? In the early 1980s researchers found partially folded forms (intermediates) of the protein structure,7,8 which have been suggested to play key roles in the folding process by simplifying the conformational search for the native state.9 Today, intermediates have been found for numerous proteins and their implications for the protein folding process have been widely discussed.10-22 In contrast, a large number of proteins have been shown to fold in a cooperative two-state process i.e. without showing any stable intermediate.23-26 Interestingly, results from kinetic measurements on variants of the apparent “two-state folder” Tendamistat indicates that partially folded meta-stable intermediates (i.e. intermediates with higher free energy than the unfolded and native states of the protein) might play an important role in the mechanism of apparent two-state folding.27

5

The mechanism by which a protein folds to its native state after biosynthesis remains one of the central unresolved issues in structural biology.14,28,29 Discovering the folding mechanism is of more than academic interest. An increasing number of different human diseases, including Alzheimer’s and Parkinson’s diseases, are caused by proteins that misfold into structures that differs from the native state, and subsequently forms aggregates in cells or the extracellular space.29,30 In order to find new therapies to treat these diseases it is highly important to identify the mechanism(s) underlying protein folding and misfolding. Conformational properties of proteins Protein stability Calculating protein stability

Under conditions that favour folding, proteins can undergo a spontaneous process whereby the unfolded protein chain folds into the native state. The difference in Gibb’s free energy (∆G) between the unfolded (U) and the native state (N) is used to describe the stability of a protein molecule. Under conditions that favour folding the overall free energy change on folding is negative i.e. the folding process is thermodynamically favoured according to scheme (1):

∆G < 0 (1)

Here kf and ku are the folding and unfolding rate constants, respectively.

In a reversible two-state system the difference in Gibb’s free energy for the unfolding process can be calculated from the ratio between the concentration of the unfolded and native states, which is equal to the ratio between the unfolding and folding constants, (equations (2) and (3)): Ku = [U]/[N] = ku/kf (2) ∆G = -RT ln Ku (3)

Here Ku is the equilibrium constant for the unfolding reaction.

kf U N

ku

6

The free energy difference can also be defined as the difference between the enthalpy (∆H) and entropy (∆S) changes according to:

∆G = ∆H - T∆S (4)

Primary forces involved in protein stability

An unfolded polypeptide makes many bonds with solvent water molecules that need to be disrupted and exchanged upon folding. Thus, new interactions within the polypeptide chain and between the chain and water molecules must be formed that can stabilize the protein molecule in the native state (i.e. the net difference in the enthalpy change must be negative to promote the folding process). In addition, the degree of freedom for the polypeptide chain will be dramatically reduced during the folding process, which will have a negative net value of ∆S, and thus a counteractive effect.

The major forces that are involved in protein stability are: hydrogen bonds, van der Waals interactions, electrostatic interactions, covalent bonds, hydrophobic interactions and conformational entropy.

Hydrogen bonds (H-bonds) are non-covalent interactions that are formed between a donor group (such as a hydroxyl -OH or an amino –NH) and an acceptor atom (such as oxygen or nitrogen).31 The average distance between the donor atom (i.e. the electronegative atom in the donor group) and the acceptor atom is around 3 Å.32 There is still an ongoing debate about how much hydrogen bonding contributes to the overall stability of the protein. Generally, however, around 70 % of all possible peptide H-bonds are formed in proteins. Thus, if the net enthalpy change per H-bond formed is -1 kcal/mol, the total enthalpy contribution for the folding reaction of a protein containing 100 peptide bonds, would be -70 kcal/mol.33

van der Waals interactions can be divided into the attractive dispersion forces, which occur between uncharged atoms due to transient dipoles that the atoms induce in each other,34 and repulsion forces, which occur when uncharged atoms come very close together without inducing dipoles. The interactions are quite weak and short ranged,31 occurring in the interior of a protein where the atoms are density packed.35 It is the large numbers of van der Waals interactions that make them important for protein stability.

Electrostatic interactions can be of three types; charge-charge (also called salt bridges), charge-dipole and dipole-dipole. Salt bridges are formed between positively and negatively charged groups and have been implicated as important stabilizing forces in protein structure (reviewed by Shortle 199236). The energetic contributions of salt bridges have been estimated to vary between 0.3 and 5 kcal/mol depending on the environment and the actual bond distance.37 However, electrostatic repulsion might, in contrast to electrostatic attractions, have a destabilizing effect on the native state of a

7

protein and may be important for the extension of the polypeptide in the unfolded state.38 Under oxidized conditions the sulfide groups of two cysteine residues can be covalently bound to each other to form a –S–S–bridge. Frequently, –S–S– bridges help to stabilize the native structure. This is because an unfolded polypeptide that contains a –S–S– bridge has a smaller number of possible conformations than an unfolded chain without the bridge. Interestingly, the protein MerP, used in this study, does not follow this trend. In this case the oxidized form is less stable relative to the reduced form by 0.7-0.8 kcal/mol. This is most probably because a negative charge at Cys17 (which is removed upon oxidation) is important for the protein’s stability.39 Covalent bonds can also occur between residues in the protein and cofactors such as metal ions. The estimated enthalpy change for Hydrophobic interactions are around –1.25 kcal/mol residue, based on the average amount of buried nonpolar area per residue in the folding process and a factor that is used to convert buried nonpolar surface to free energy.40,41 During the folding process of water-soluble proteins the hydrophobic residues become buried within the molecule and the hydrophilic parts are exposed to the water. This increases the entropy of the system since the surrounding water molecules become less ordered when they are not in contact with non-polar groups. This gain in entropy is thought to be a very important force promoting the folding reaction and is described as the hydrophobic effect. In contrast, the conformal entropy of the polypeptide chain will be greatly reduced during the folding process, and this is the major destabilizing force42 which needs to be balanced by the hydrophobic effect combined with the previously described interactions. Thus, most folded proteins are only marginally stable where the free energy of unfolding ranges between 3-15 kcal/mol.43 Protein folding The Levinthal paradox In the beginning of the 1960s Anfinsen found that proteins can fold reversibly into thermodynamically stable states.44 This came to be called thermodynamic control and was taken as evidence that the folding of proteins to their native state is “pathway independent” (i.e. the native structure is determined only by the final native conditions). However, this process should take a long time because it may require an extensive search for the unfolded polypeptide to reach its native state. In 1969 Levinthal45 pointed out that random searching by a polypeptide to achieve its functional

8

fold would require immense amounts of time, even for small, single-domain proteins because there are so many possibly conformations. Assuming that each of the backbone φ (phi) and ψ (psi) angles (the angles of rotation about the bond between the N and the α-carbon and between the α-carbon and carbonyl-carbon, respectively) in a polypeptide chain can adopt only three different values, then a protein containing 100-residues can adopt ca 3200 ≈ 1095 different possible conformations. If it takes around 1 ns for the transition between these conformations to occur the protein would need 1078 years to sample all these structures in a random search to find its native conformation.46

But proteins need to fold quickly, in a biological timescale, usually in the order of milliseconds to minutes. This came to be called “kinetic control”, implying that folding happens quickly because it is “pathway dependent”.44 This raise questions about how protein folding could be simultaneously pathway dependent and pathway independent.47 These two conflicting considerations constitute the “Levinthal paradox”. The “classical view” and the “new view” of protein folding

There are two major different views on how this paradox may be resolved20: (1) the “classical view” and (2) the “new view”.

In the “classical view” (see Figure 4) it is proposed that the paradox can be resolved by the existence of partially unfolded intermediates since they might allow folding to occur sequentially step by step, which would probably reduce the scale of the conformational search. The search for folding intermediates (I) that occur in a sequence from the unfolded form (U) to the native form (N) has been thought to provide opportunities to study conformational changes in the protein structure during the folding process.46

Figure 4. The ”classical view” of protein folding.

U I1 I2 ⋅⋅⋅ N

Unfolded state Intermediates Native state

9

In the “new view” (primarily based on theoretical studies of protein folding) the paradox is resolved by describing the folding process in terms of a funnel-shaped energy landscape. The funnel can be smooth if conformational dissimilarities with the native state is accompanied with a large difference in the free energy48 (see Figure 5a). On the other hand, in a rugged energy landscape, small changes in the free energy can be accompanied with large conformational changes (see Figure 5b).44 In this model there is no specific pathway for the folding process. Instead, the unfolded state constitutes an ensemble of polypeptide chains from which each molecule follows its own route, and other chains follow different trajectories, to the native state.20,44,47

Figure 5. The ”new view” of protein folding. (a) A smooth funnel. (b) A rugged funnel. Created by Jon Tibell. Intermediates

For several proteins intermediates have been found that are stable enough to be populated under certain conditions. Many proteins form a “molten globule” intermediate that is quite compact and has a native-like secondary structure, but only a few well-organized tertiary structure elements.49 Many studies support the idea that these “molten globules” represent intermediates that are on the protein folding pathway because they can have a very native-like structure, supporting the “classical view”.17 In the “new view” intermediates are not necessary for the unfolded polypeptide chain to fold to the native state. Instead, these intermediates might occur as a result of the ruggedness of the energy landscape (see Figure 5b).20 This can be visualized as skiers going down a mountain who may have to traverse valleys at altitudes higher than the lowest point on the mountain range before they reach the deepest valley.44

N N

U U U U U U U U U U U U U U

a b

10

Two-state folding mechanism Intermediates with characteristics of the molten globule state are

normally found for proteins that are relatively large (> 100 amino acids).49-51 However, numerous small, single-domain proteins, with 100 amino acids or less, have been found to fold without any accumulation of detectable intermediates.52 This means that proteins can fold quickly without the guidance of intermediate conformations. These proteins are often said to fold in a two-state mechanism i.e. only the folded and unfolded forms of the protein are populated during the folding event. Some following criteria must be fulfilled for a protein to be classified as a “two-state folder”.53 First, the protein must exhibit single-exponential decay in both the forward (folding) and the reverse (unfolding) direction with no burst-phase within the dead time of the experiment i.e. the total difference in signal between the native and unfolded states must be detected in the kinetic experiment. Second, the ratio of the forward to reverse rate constants (kf/ku) must equal the equilibrium constant Kf. Third, the same folding behaviour should be found when the folding process is measured by different techniques that probe different parts of the protein structure.

Two different concepts have been postulated for the two-state folding behaviour of proteins: the transition-state theory and the funnel model. In the transition-state theory (see Figure 6) an energy barrier impedes, and slows, the transition of the unfolded polypeptide towards a diffuse structure called the transition state (TS). After reaching TS there is a rapid transition to the final native conformation. In this model U, TS and N are macrostates, each of which contains a distinct ensemble of microstates (specific atomic structures or chain-conformations) that occur in a sequence along a single reaction coordinate i.e. U TS N.53 The single-exponential behaviour arises from the formation of TS, which is the bottleneck in the protein folding event.53

Figure 6. The transition-state theory of two-state folding.

Reaction coordinate

Macrostates that contains distinct ensembles of microstates

N U

∆G TS

11

In the funnel model (derived from microscopic theory and simulations) of protein folding the two-state kinetics can be described by a very smooth energy landscape, Figure 5a, in which the funnel is flat at the top and steeper and deeper toward the native structure.44 The funnel model shows how a single native structure can be reached rapidly from a large number of different unfolded conformations. However, a smooth funnel lacks a barrier and, thus, while the funnel model can explain the speed of folding it does not provide a good explanation of two-state kinetics.53 In contrast, the TS theory can explain the single-exponent behaviour, by postulating the existence of an energy barrier, but not really the folding speed because a barrier will slow rather than accelerate the process. So the question remains, how can we reconcile these conflicting behavioural patterns?

Recently, Schonbrun and Dill presented three conclusions about the microscopic basis for fast-folding kinetics.53 They suggest that fast folding results from the occurrence of parallel microscopic routes. In the beginning of the folding process the polypeptide does not spend any time searching in the unfolded conformational space, rather the protein molecules “fall down” the folding funnel like water down a water-fall. The bottleneck, which gives rise to the single-exponential behaviour, is found where the funnel becomes narrow due to the reduction in the large numbers of folding trajectories down the funnel landscape. Initially, folding happens very quickly because the unfolded conformation has a much higher probability of making new native contacts to proceed downhill, simply because it can happen in so many different ways, i.e. there are many routes downhill. In contrast, it takes much longer for near-native conformations to make new native contacts because there are fewer ways of doing it, i.e. there are only a few routes downhill.

Hidden intermediates

Recent studies on proteins that show apparent two-state folding mechanisms have revealed the existence of “hidden” intermediates (HI) that occur sequentially along the folding pathway,20 but are masked by the transition state. Hidden intermediates cannot be detected by conventional kinetic methods because they exist at energy levels that are either higher than the unfolded state or located at the native side of the rate-limiting energy barrier. The findings of HI have been invoked to question the validity of the funnel model for two-state behaviour because, in contrast to a folding pathway with a sequence of kinetic species as illustrated in Figure 4, the funnel model suggests that multiple folding events occur in parallel, there is a continuum of different conformations rather than a discrete hierarchy and the energy surface is most likely smooth i.e. without any intermediate states.54

12

However, in computer simulations using a two-dimensional folding model Ozkan et al54 have demonstrated that hidden intermediates and hidden TS can occur in parallel rather than along a single reaction coordinate. Consequently, even for apparent two-state proteins the shape of the energy landscape might be a highly complex funnel with a certain degree of ruggedness. Structural mechanism of protein folding

In searches by scientists over the past 40 years to find a model describing how the structural parts in the protein structure are formed and added during the folding process, different mechanisms have been proposed, a few of which are discussed here.

In the nucleation-growth model55 one or more well-defined kinetic nuclei, containing secondary structure elements, are first formed from which the tertiary structure is propagated.

Two other, very similar models are the framework model56 and the diffusion-collision model57, which are based on the idea that the hydrogen bonded secondary structure is the first to be formed. The pre-formed secondary structure elements are then added, either by propagation of the structure in a stepwise manner or by diffusion and collision, to finally form the tertiary structure in the native state.

In the hydrophobic collapse model58 folding is thought to be driven by water molecules being squeezed out from the hydrophobic parts of the polypeptide chain. This will cause the chain to collapse into a state from which secondary elements can start to grow.

According to the nucleation-condensation model13,59 the folding process starts with a rapid random search of conformations in the unfolded state, and native-like secondary structures in the unfolded polypeptide become stabilized by long-range interactions as folding proceeds. When sufficient interactions have been made the transition state is reached and rapid formation of the final structure ensues. If conformations are formed with non-native interactions, which have to be disrupted before the transition state is reached, they will slow down folding because of the loss of their stabilization energy.60 In contrast to the nucleation-growth model the folding nucleus of this model is rather diffuse. Is there a universal model for protein folding?

A few years ago it was shown by Baker and coworkers24,61 that the folding speed of two-state proteins is related to the contact order (CO) of the protein, i.e. the folding speed is related to the average sequence distance between all pairs of contacting residues normalized by the total sequence length (a high CO value reflects long distances between the contacts in the

13

polypeptide). This indicates that proteins that exhibit more sequence-local contacts tend to fold more rapidly than proteins that are stabilized to a larger degree by long range contacts. In a very recent study Kamagata et al50 proposed a unified mechanism of protein folding for “two-state folders” and proteins that fold through the formation of an intermediate i.e. “non-two-state folders”. They found that the rate constants for the formation of folding intermediates, and formation of the native state from intermediates, reflect the native state topology in a similar way to that found for “two-state folders”. They concluded that the folding mechanism of “non-two-state folders” may represent a general model of protein folding. In this model, two-state folding appears to be a truncated version of the folding mechanism found for “non-two-state folders”.

In a comparative study of transition states, Nötling and coworkers62 found that the nucleation-condensation model seems to be valid for five of six small proteins they tested. Later, in a recent paper by Daggett et al63 it was proposed that variations of the nucleation-condensation model (see the description of the model above) can be used to describe the overall features of folding of most protein domains. If the propensity for stable secondary elements increases in the folding nucleus, folding can occur in a more hierarchical manner, thus following the framework model. However, if hydrophobic interactions are formed too rapidly and strongly the protein might fall into a molten globule “trap” according to the hydrophobic collapse model.

The question still remains if there is any way of combining the “classical view” with the “new view” of protein folding. A cross-section of the funnel landscape, see solid line in Figure 7a, shows that the conformational entropy (∆S) of the polypeptide chain decreases much faster than the internal energy (∆H) of the system (the broken line illustrates a good match between ∆S and ∆H). This mismatch causes an energy barrier that is visualised in transition-state theory (see Figure 7b). A “bump” in the reduction of ∆S promotes the formation of an intermediate (I) (dotted line in Figure 7a,b) between the native and unfolded states in the energy diagram (Figure 7b). However, in contrast to transition state theory, the funnel model clearly shows that there might be more than one reaction coordinate for the protein molecule.

If a protein molecule has an almost infinite number of folding pathways (i.e. all kinds of interactions that occur in the unfolded peptide favour the folding process) the folding would be a highly cooperative process that occurs rapidly. This may be the case for proteins that fold extremely quickly, with a time constant of a few ms. One example might be the protein CspB, which folds in less than 1 ms.64 In contrast, apparent two-state proteins that fold more slowly (with a time constant of hundreds of ms) might need an extensive search in the unfolded conformational space before

14

the right interactions are found that will promote the folding process. There might be several routes for the folding process, requiring a complex search for the polypeptide to find the interactions that must be formed before folding can be completed. It is the search for these interactions that takes time. This might be the case for the protein MerP (see Papers II and III), which takes 400 ms (= the half-time for the folding process) to fold to the native state.

Figure 7. (a) A cross-section of the funnel landscape. (b) A energy diagram of protein folding. Protein unfolding Dynamics of proteins

The marginal stability of a protein has important consequences. In order to perform its function the structure needs to be sufficiently flexible. Furthermore, at appropriate times proteins need to be degraded in the cell, and for this too the protein structure must not be too stable. Thus, in many cases proteins can be easily denatured (unfolded) by changing their physical or chemical environments.43 The most common methods for doing this experimentally are by increasing the temperature or by adding a chemical denaturant such as urea or GuHCl.

∆S

N

U U U U U

Internal energy

I

I N

I U

∆G

Reaction coordinate

mismatch

a b

15

Thermal denaturation In thermally induced protein unfolding the enthalpy and entropy changes are dependent on the temperature, because of the differences in heat capacities between the folded and unfolded forms of the protein (∆Cp) according to65:

PCT

H ∆=∂∆∂

)(

)( (5)

and

T

C

T

S P∆=

∂∆∂

)(

)( (6)

The change in heat capacity reflects the fact that more heat is required to raise the temperature of a solution of unfolded protein than a solution of folded protein of the same concentration.41 ∆Cp is related to the hydrophobic contribution to protein stability, and in general to the change in solvent-accessible hydrophobic surface associated with the unfolding transition.66

When integrating equations (5) and (6) one can determine the enthalpy difference between the native and unfolded states over a temperature range (TR – T) in which ∆Cp is known65 according to:

dTTCTHTHT

T

PR

R

)()()( ∫∆+∆=∆ (7)

and

dTTT

CTSTS

T

T

pR

R

)()()( ∫∆

+∆=∆ (8)

where TR is a reference temperature. If ∆Cp is constant the following equations will apply:

∆H = ∆HR + ∆CP (T-TR) (9) ∆S = ∆SR + ∆CP ln (T/TR) (10) From equation (4) it can be concluded that at the midpoint melting

temperature of the temperature-induced unfolding curve (Tm) (i.e. the temperature where ∆G is equal to zero) the following relationship is valid:

16

∆Hm = Tm∆Sm (11) and thus:

∆S = ∆HR/TR + ∆CP ln (T/TR) (12)

By inserting equations (9) and (12) into equation (4) and using Tm as the reference temperature, a thermal unfolding curve can be generated according to the Gibbs-Helmholtz equation:

∆G = ∆Hm (1-T/Tm) - ∆CP [(Tm-T) + T ln (T/Tm)] (13) Equation (13) was used in Paper II to calculate the thermodynamic parameters for individual amide protons from the hydrogen exchange data. Chemically-induced denaturation

The most common chemical agents that are used to unfold proteins are GuHCl and urea. It is not yet clearly understood how these chemicals act to unfold proteins. Either the protein unfolds via direct binding of the chemicals to the polypeptide chain stabilizing the unfolded state (i.e. via the increase in solubility of the parts of the protein that are exposed upon unfolding)67,68 or the protein unfolds via an indirect mechanism by perturbation of the water. According to the latter model the denaturants increase the mobility of the water molecules, and thus increase the entropy of the water. If so, protein folding may be strongly affected by the entropy of the water. In general, features of both of these models may be involved in the unfolding event. From a chemically-induced unfolding curve the equilibrium constant for the unfolding process can be calculated in the transition region where the free energy change is close to zero (i.e. the unfolded and folded forms of the protein are almost equally populated according to equation (2)). The free energy change in this region can then be calculated using equation (3). From a plot of ∆G versus the chemical concentration the stability of the protein, in the absence of denaturant, can be calculated by extrapolating the free energy from the transition region back to zero concentration of denaturant using equation (14)69: ∆G ([denaturant]) = ∆G(H2O) – m [denaturant] (14) where ∆G(H2O) is the free energy change at zero concentration of denaturant, while m describes how ∆G is related to the denaturant concentration and is a measure of the new surface area exposed to solvent

17

upon unfolding. In Paper III equation (14) was used to fit the plot of the free energy for individual amide protons versus different concentrations of GuHCl. However, this method assumes that there is a linear dependence of the free energy on the denaturant concentration, and this might not be the case for all proteins. When using GuHCl the free energy curve can show a pronounced upward curvature according to the denaturant binding model of Tanford.70,71 In this model the degree of binding of denaturant to a limited number of sites of the protein molecule is considered. If the sites are equivalent and non-interacting then equation (15) becomes valid: ∆G([denaturant]) = ∆G(H2O) - ∆nRT ln (1+Kb[denaturant]) (15) Here ∆n is the number of additional denaturant binding sites, with averaged binding constant Kb, that are exposed when the protein unfolds. Equation (15) was used in Paper III to fit the hydrogen exchange data that showed an upward curvature in the plot of the free energy versus [GuHCl]. Protein misfolding

Several different human diseases are now known to be associated with aggregations of specific proteins. These aggregates are thought to be directly or indirectly involved in the pathological conditions associated with the disease. The final form has a well-defined fibrillar structure that is known as amyloid, and thus the term amyloidosis is frequently used to describe the clinical conditions with which they are connected.30

Although the molecular structure of amyloid fibrils at atomic resolution is not known, a general structure has been proposed consisting of parallel or anti-parallel β-sheets that are oriented perpendicular to the direction of the fibril axis. This structural concept is based on fibre-diffraction,72 cryo-electron microscopy,73 mass-spectrometry74 and solid state NMR spectroscopy75 studies.

Today an increasing number of proteins that are not involved in amyloid diseases have been reported to have the ability to form fibrils76-78 including the model protein MerP examined in this study (see Figure 8). Thus, although only a handful of proteins are involved in amyloid-related diseases (~25) it seems that the ability of proteins to form fibrils might be a generic property of the polypeptide chain.30

The exact mechanism causing the series of events that give rise to amyloid formation is not yet completely understood. However, both biophysical studies of in vitro aggregation and animal model studies support the idea that ordered pre-fibrillar oligomers or protofibrils may be

18

responsible for cell death rather than the fibrillar end-product of the protein aggregation pathway.79-83

Figure 8. Fibrils of MerP. Techniques used in this study Nuclear Magnetic Resonance Basic background Nuclear magnetic resonance (NMR) occurs when the nuclei of certain atoms are immersed in a static magnetic field and exposed to electromagnetic radiation. Nuclei that have a nuclear spin (I) different from zero behave as if they were spinning around their own axis. When nuclei with I = ½, such as protons and 15N nuclei, are placed in a magnetic field, each nuclear spin aligns such that its magnetic moment is either in the same direction as the applied field, or directly opposed to it, with the former alignment occupying a lower energy level. Boltzmann statistics can be used to predict the relative populations of the two energy levels.

19

e Tk

E

B

N

N ∆−=α

β (16)

where Nβ and Nα are the number of nuclei in the upper and lower energy levels, respectively, kB is the Boltzmann constant and ∆E is the difference in energy between the α and β states. ∆E is dependent on the magnetic flux density, the specific nucleus and the surrounding conditions. The populations of the energy levels are nearly equal, however the lower energy level population exceeds that of the upper level in the order of parts per million (ppm).84 One-dimensional NMR The spin in the lower energy level can be excited to the higher energy level by the absorption of energy from electromagnetic radiation at a specific resonance frequency, in the radio frequency (RF) region of the spectrum, according to: Ehv ∆= (17) where ν is the resonance frequency. A 1D NMR spectrum is achieved by simultaneously exciting all the nuclei of one species in a sample using a short radiofrequency pulse. Immediately after the nuclei have been excited the data acquisition phase follows, where the free induction decay (FID) from excited spins is monitored as they relax back to the ground state. This results in a time domain signal which, after Fourier transformation, gives a frequency domain spectrum.84 For convenience, the frequency is converted to the chemical shift (δ), which is the frequency difference between the sample and an internal standard divided by the sample frequency and expressed in ppm. The signal intensity is proportional to the population difference between Nβ and Nα, and is thus proportional to the concentration of the sample. In a 1D proton NMR spectrum of a protein molecule there will be several peaks at different chemical shifts since the shifts are very sensitive to the molecular environment of the hydrogen atoms in the molecule. However, extensive spectral overlaps will occur between the signals, which complicate interpretation of the spectrum. To overcome this problem multi-dimensional NMR can be used, which can greatly improve the resolution of the peaks because in this type of analysis the shifts are separated in more than one dimension, reducing the spectral overlap of the signals.

20

Two-dimensional NMR Multi-dimensional NMR is routinely used nowadays to study protein molecules. The basic principles of two-dimensional NMR experiment are to insert a delay after the excitation of the nuclei and before the data acquisition phase, which is varied between different NMR analyses. The second dimension of the spectrum is obtained by performing a Fourier transformation with respect to the inserted time domain. Using this technique it is possible to disturb the spin of one nucleus and detect its effect on the spin state of another nucleus since they are coupled with each other. In a COSY spectrum, spin effects from adjacent nuclei transferred through covalent bonds are analysed. This approach allows separate spin-systems to be identified, and thus atoms that are close in sequence can be identified and their chemical shifts can be assigned.85 Assignment of individual amide protons is needed when studying the exchange rates of individual amide protons with surrounding molecules (Papers II and III). In a NOESY experiment spin effects that are transferred between nuclei through the space are analysed, thus allowing the distance in space between adjacent atoms to be determined. This type of experiment can be used to solve the structure of proteins.85

By labelling proteins with 15N it is possible to analyse them in 1H-15N Heteronuclear Single Quantum Correlation (HSQC) experiments,86 in which the spin interactions between the proton and the nitrogen in the amide group of residues can be monitored. Each amino acid residue (except proline) gives rise to a cross-peak in the NMR spectrum where the chemical shift for the amide nitrogen is displayed along the vertical axis and the chemical shift for the amide proton is shown along the horizontal axis. The 1H-15N HSQC spectrum of MerP displays a good resolution of the signals (see Figure 9) and since the chemical shift is sensitive to the environment of the peptide group NH this approach was used in the analyses reported in Paper IV to detect conformational changes of MerP at different GuHCl concentrations. The proportional relationship between the signal intensities and the concentration of amide protons in the sample provides the basis for using NMR to monitor exchange rates of the amide protons with surrounding molecules, and this was used in Papers II and III to explore the dynamics of the protein structure at different amide positions in MerP. This technique is called hydrogen exchange analysis, and will be described in the following section.

21

Figure 9. 1H-15N HSQC spectrum of MerP at pH 6.5. Hydrogen exchange Basic background The hydrogen exchange technique was conceived by Kaj Linderstrøm-Lang in the early 1950s.87 Based on the idea that the amide proton in the peptide group constantly exchanges with the hydrogen of the surrounding water molecules he inferred that the exchange rate of the amide proton can be correlated with the accessibility of the proton of the water molecules. Further, this accessibility must depend on the dynamics of the protein structure at the specific proton site. Amide hydrogen-deuterium exchange

11

50

65

61

58

13

53

26

17,22

37 68 4

52 9

2

16

23

55

66

48

41

14

40

25 42 20

57

56 60

19 27

64

33 28

21 24

31

12

51

34

6 70

36

54

39 63

5 43

32 47

7

45 49

71 29

72

59 8

44 46 38

15 3 35

30

62

69

9.5 9.0 8.5 8.0 7.5 7.0 6.5

105

110

115

120

125

130

1H

15N

ppm

ppm

22

(1H/2H) is currently one of the most promising techniques for exploring the energy landscape of protein folding.88 Experimentally the exchange rate (kex) of the amide proton can be measured by dissolving a lyophilised 15N-labelled protein in 2H2O and examining it in a couple of 1H-15N HSQC experiments at different time points. Since 2H is silent in 1H NMR analyses the rate of exchange between 1H-15N and 2H-15N can be determined by fitting the decay in the cross-peak volumes versus time plots to a single exponential decay (see Figure 10).

0

20000

40000

60000

80000

100000

120000

140000

0 20 40 60 80 100 120

Figure 10. Single exponential decay of cross-peak volumes of an amide proton in MerP (Paper II). The exchange rates can then be analysed, using a two-state model, according to scheme (18)89,90: kop krc kcl N1Hclosed ↔ N1Hopen → N2Hopen ↔ N2Hclosed (18) kcl kop According to this model the exchange of NH is in equilibrium between a solvent-inaccessible conformation (NHclosed) and a solvent-accessible conformation (NHopen),

91 where kop and kcl are the opening and closing rate constants between these two states, respectively. The rate constant for the exchange of the amide proton in the NHopen conformation is denoted krc. This rate constant depends on adjacent side chains, temperature and pH and can be theoretically calculated under different conditions by using

Time (min)

23

parameters obtained from experimental studies on model peptides.92 At pH below 3 the catalysis of krc is dominated by H3O

+ and at pH above 3 OH-

dominates, resulting in rates ranging from about 10-3 s-1 close to pH 3 to greater than 103 s-1 above pH 9, at 25 °C.93 Using the two-state model the following equation for kex has been derived94:

( )

++−−

++=2

411

)(

2/1

2rcclop

rcop

rcclopex

kkk

kk

kkkk (19)

Under native conditions the equilibrium constant for the opening reaction ranges approximately between 10-2 and 10-11, given a typical value for the free energy change between 3-15 kcal/mol at 25 °C.91 Thus, under these conditions kop << kcl and equation (19) can be simplified as follows:

rccl

rcopex kk

kkk

+≅ (20)

In the so-called EX1 limit, kcl << krc and thus equation (20) can be reduced to equation (21): opex kk = (21)

Under this condition the exchange rate is a direct measurement of the rate constant for the opening reaction according to scheme (18) and thus the kinetics of the protein unfolding/folding reaction can be studied.95 However, for most native proteins at room temperature and pH ≤ 7 the exchange occurs via an EX2 mechanism96,97 where kcl >> krc. In this limit case equation (20) can be reduced to equation (22):

rccl

opex k

k

kk

= (22)

24

From equation (22) the equilibrium constant (Kop) for the opening reaction that leads to the exposure of the amide proton can be calculated according to:

rc

ex

cl

opop k

k

k

kK == (23)

Thus, the free energy changes for the opening reaction that promote exchange of the amide proton can be calculated using equations (3) and (23), as in Papers II and III for individual amide protons at different temperatures and GuHCl concentrations, respectively.

If the amide proton is buried and inaccessible to the surrounding molecules, or if the proton participates in a hydrogen bond, the protein structure around the proton needs to unfold before exchange can take place. This can happen in any one of three ways87 as shown in Figure 11. In the first case (Figure 11a) the proton requires a global unfolding event, which will expose all protons in the molecule, to become accessible. In the second case (Figure 11b) only part of the protein molecule needs to be unfolded prior to exchange, and thus only a partial unfolding event is required for the exchange to take place, and finally the proton can exchange by a local unfolding event that involves very little exposure of the folded surface in the protein (Figure 11c).

Figure 11. Three unfolding models for protein hydrogen exchange; (a) global unfolding, (b) partial unfolding and (c) local fluctuation.

a

b

c

25

Implications from hydrogen exchange studies The most common way to study hydrogen exchange is to measure the exchange rate at different temperature and/or denaturant concentrations in the pre-transitional region of the equilibrium curve (i.e. the region before the protein starts to unfold). This is called native-state HX98 and has been used in combination with NMR to study the structural stability of a wide range of different proteins at amino acid residue level.99-113 The results from these studies have greatly increased knowledge of protein folding, and indicated that the folding process can differ substantially between different proteins.

Using this method partly unfolded forms (i.e. intermediates) have been found for proteins that fold by an apparent two-state mechanism,100,106-108,110 implying that the folding landscape for these proteins might not be a very smooth surface. For two of these proteins, cytochrome c (cyt c)100 and Rd-apocyt b562,

108 the partially unfolded forms belong to substructures with a well-defined energy and, interestingly, for cyt c the partially unfolded forms resemble kinetic intermediates found on the folding pathway,114 indicating that the folding occurs sequentially in a step-wise manner.20 The question then arises, is it possible to study kinetic events on the folding pathway by using native state HX? The answer seems to be affirmative for some proteins, since there is a connection between amide protons that exchange most slowly and folding nucleation sites in some cases.99,108,115-119 However, no direct relationship between hydrogen-exchange behaviour and the protein folding pathway has been found for barnase,101,120,121 CI2105,122 or human acidic fibroblast growth factor 1 hFGF-1,123 so the issue remains to be fully resolved. Another interesting feature that has been observed for T4 lysozyme,107 hFGF-1110 LysN106 and MerP (Papers II and III) is that they all have broad distributions of free energies that are widely dispersed throughout the structure. The observation that residues with similar energies do not cluster in well-defined areas suggests that unfolding does not occur via the formation of a few discrete partially unfolded forms. Instead, it is consistent with the occurrence of a large number of energy levels between the native and unfolded forms. This feature might reflect the existence of several partly unfolded conformations on different folding trajectories that occur in parallel, in line with the “new view” of protein folding. Arrhenius analysis of hydrogen exchange data Expressing equation (22) in energy terms using the Arrhenius relationship gives the following equation48:

RT1

‡E

RT

G

ex eeCkop −∆−

= (24)

26

where ∆Gop is the free energy of opening the native state, E‡ is the activation energy for exchange from an unstructured peptide and C1 is a constant. Equation (24) can be rearranged by taking the natural logarithms of both sides to:

RT

lnln‡

1

E

RT

GCk op

ex −∆

−= (25)

By expressing the free energy in terms of enthalpy and entropy according to equations (4), equation (25) can be rewritten as:

RT

lnln‡

1

E

R

S

RT

HCk opop

ex −∆

+∆

−= (26)

When taking the differential of lnkex with respect to 1/T the following equation is obtained:

)(1

R1ln ‡

‡

EHR

E

R

H

T

kop

opex +∆−=−∆

−=∂

∂ (27)

The slope from an Arrhenius plot of the exchange data (i.e. the natural logarithms of the exchange rates versus 1/T) is equal to the expressions in equation (27), resulting in the following relationship:

R

EEH

Ra

op −=+∆− )(1 ‡ (28)

where Ea is obtained from the slope of the Arrhenius plot and E‡ can be derived, following Bai et al92 from model peptide studies. Thus, the enthalpy difference between NHopen and NHclosed (∆Hop) can be calculated from the following equation: ∆Hop = Ea- E

‡ (29) The enthalpy difference between NHopen and NHclosed for amide protons that exchange by local fluctuation was calculated and compared with ∆Hm in Paper II.

27

Fluorescence spectroscopy

The emission of light of a longer wavelength by some atoms and molecules as they relax back to the ground state following excitation by radiation at a particular wavelength is called fluorescence. Fluorescence emission from proteins originates from the aromatic amino acids124 and is dependent on the molecular environment and mobility of the fluorophores. Thus, this technique can be used to study the conformational changes that occur during unfolding/refolding of proteins. Among the three aromatic amino acids found in proteins (phenylalanine, tyrosine and tryptophan) tryptophan is the most sensitive fluorophore, but tyrosine can also be used if no tryptophan is present in the protein sequence. The fluorescence properties of proteins are highly specific, and their behaviour on unfolding is not always predictable.125 However, the emission maximum of an unfolded protein is usually shifted to a longer wavelength relative to the folded protein, due to the transfer of the fluorophore from a hydrophobic to a hydrophilic environment.125 The changes in fluorescence emission upon unfolding are strongly related to the conformational changes in the tertiary structure of the protein. Thus, conformational changes of the tertiary structure can be monitored by measuring the fluorescence of a protein sample at various concentrations of denaturant. From the resulting data the conformational stability of the protein can be calculated (Paper I). Structural changes can also be measured kinetically by fluorescence analyses, and from the resulting data the rate constants for the unfolding/refolding events can be calculated (Figure 5 in Paper I). Circular Dichroism spectroscopy When a chromophore is part of an asymmetric structure, left-handed and right-handed circular polarized light are absorbed to differing extents.124 This phenomenon is called circular dichroism (CD). In the far-UV spectrum (below 250 nm) peptide groups located in asymmetric protein structures of the native protein (such as α-helices, β-sheet and turns) absorb the two light-forms differently. Far-UV CD gives a characteristic CD spectrum for folded proteins (depending on the content of the secondary structure) which disappears when the protein becomes unfolded because of the loss of symmetry in the unfolded state (Figure 2c in Paper I). Thus, far-UV CD can be used to probe structural differences in secondary structure under different folding conditions. In the near-UV spectrum (between 250 and 320 nm) the major chromophores are the side chains of the aromatic amino acids and therefore the fine structure in a near-UV CD spectrum of a folded protein reflects the

28

asymmetry in the tertiary structure. This fine structure is not present in the unfolded polypeptide because the aromatic side chains are in a symmetric environment in the unfolded state (Figure 2b in Paper I). Thus, near-UV CD can be used to monitor structural changes in the tertiary structure under different folding conditions. Multivariate Data Analysis Paper IV describes a method for using multivariate data analysis (MVDA) to evaluate complex data sets that contain large amounts of information. In general MVDA can be used for three main purposes126: (i) to obtain an overview of the data table, (ii) to classify groups of observations, and (iii) to find the relationship between two block of data, X and Y, which can be used to predict Y from X for new observations. Problems of the first two kinds can be solved by principal component analysis (PCA) and problems of the third kind by partial least square projections to latent structures (PLS).126 PCA is designed to extract and display the systematic variation in a data matrix (X). Such matrices contain N rows (observations) and K columns (variables). The matrix in Paper IV has 11 rows (each of which corresponds to a specific GuHCl concentration) and 204 800 columns containing the free induction decay data obtained from the 2D NMR measurements recorded at each GuHCl concentration. In a multidimensional space each observation in the data matrix (11 in Paper IV) defines a point in the space spanned by the K variables (204 800 in Paper IV). Thus, the X-matrix in paper IV corresponds to a swarm of 11 points in a space containing 204 800 vectors. The point-swarm can be described using a mathematical model which approximates the data. This involves computing the first principal component (PC1): the line in the multidimensional space that best describes the variance in the data. Each observation can be projected onto this line to obtain a co-ordinate value that is known as the score value. Usually at least a second principal component (PC2), which is orthogonal to the first PC, is needed to model the systematic variation of the data set adequately. In Paper IV the first three principal components of the PCA analysis were calculated resulting in an informative 3-dimensional score plot (Figure 2 in Paper IV). This plot shows an overview of the variation in the NMR data and made it possible to divide the 11 NMR measurements into three sub-groups. PLS is a regression extension of PCA that can be used to relate two data matrices, X (factors) and Y (responses), to each other. The X-matrix consists of a set of observations (rows) and variables (columns) also referred to as independent variables. The Y-matrix has the same number of observations but another set of variables, sometimes referred to as

29

dependent variables. In PLS each observation corresponds to two points (rather than one in PCA). One point is located in the X-space and the other in the Y-space. In Paper IV each sub-group (from the PCA analysis) was subjected to PLS in order to detect spectral changes between the different NMR measurements within the group and the way these changes are related to the GuHCl concentration. To do this, the denaturant concentration was used as the variable in the Y-matrix and the free induction decays obtained from the NMR measurements were used as variables in the X-matrix. Each row in the X- and Y-matrix corresponds to a specific GuHCl concentration. The first PLS component is the line in the X-space which best describes the variation in the point-swarm and provides a good correlation with the y-vector (with only one y-variable, the Y-space reduces to a one-dimensional vector). Each observation can be projected onto this line to acquire a score value in a similar way as in PCA. The first X-score vector t1 is the scores of all the observations (11 in Paper IV). For each PLS component t are linear combinations of X with weights w where X is the data matrix and w is the weight vector that shows how the variables in the X-matrix contribute to the PLS-model. Thus, w describes how the variables in the X-matrix contribute to the direction in the multidimensional space that best describes the variation in the point-swarm of the X-space and provides a good correlation with the y-vector. In Paper IV the first PLS weight vector, w1 was subjected to Fourier transformation, resulting in an NMR spectrum for each sub-group reflecting the changes between the various recorded NMR spectra induced by the denaturant (Figure 3a-c in Paper IV). The protein MerP The function of MerP has been studied by Sahlman and coworkers who found it to be involved in the detoxification of mercury ions.127 The mer structural genes, which provide resistance to Hg2+, are located on transposon Tn21 in Plasmid R100: the initial antibiotic-resistance plasmid that appeared in 1956 in a Shigella flexneri strain in Japan. MerP is a water-soluble protein located in the periplasm of the bacteria and has the capacity to bind Hg2+ to two thiol groups (Cys14 and Cys17) in its reduced form.39 The protein contains 72 amino acids after removal of the periplasmic signal sequence. High-resolution solution structures of oxidised MerP determined by NMR128,129 show that the secondary structure comprises four anti-parallel β-strands and two α-helices on one side of the sheet in the order β1α1β2β3α2β4 (see Figure 3). This topology is identical to that of three other proteins (U1A, S6, ADA2h) and very similar to the topologies of two

30

others, AcP and HPr, for which some folding properties have been characterized.130-135 In the studies underlying this thesis the oxidized form of the protein was examined, and no folding studies have previously been published using MerP as a model protein. The oxidized form was chosen because it is easier to keep the protein in a homogenous form in this state and thus control the system. Furthermore, only one disulphide bridge is formed between two adjacent amino acids (Cys14 and Cys17) that were known to be located in a loop region128 and thus might not have a major effect on the overall structure or folding behaviour of the protein. Summary of the papers I. Fluorescence and CD analysis of the folding properties of MerP The first equilibrium studies of MerP using fluorescence and CD indicated that the folding process could be fully reversed by manipulating pH, temperature or GuHCl and that it followed a two-state model in which only the native and unfolded conformations are populated. Interestingly, fluorescence-based kinetic studies of the folding event showed that the rate constant for the major fast-folding population is very slow for such a small protein (kf = 1.2 s-1). In comparison, U1A136 and ADA2h137 fold 280 and 630 times faster, respectively, despite having identical topology to MerP. Furthermore, the kinetic measurements of the unfolding/folding events showed that the unfolded population of MerP is inhomogeneous. The stability of the protein (∆G(H2O) from equation (14)) was found to be 2.6-4.0 kcal/mol at 25 °C, depending on the pH, implying that approximately 0.1 to 1.2 % of the protein molecules are unfolded under folding conditions. The corresponding m value from equation (14) was found to be around 2.4 kcal mol-1 M-1, which could be expected for a protein of this size.138 Thus, a major challenge was to try to understand the underlying reasons for these folding properties of MerP. The intrinsic fluorescence from Tyr66 (the only fluorophore in MerP) increases markedly upon denaturation in GuHCl and the spectrum is shifted to the blue region. Clearly, therefore, Tyr66 is more quenched in the native state relative to the unfolded state due to the lower intensities seen in the folded spectrum. Another interesting observation is the blue shift of the spectrum upon denaturation, indicating that the atoms surrounding Tyr66 in the native state cause relaxation around the excited state dipole leading to a lowering of the energy in the native state relative to the unfolded state. Usually, if the fluorophore is located in a hydrophobic region in the native

31

state the spectrum will be shifted to red upon denaturation because of the relaxation caused by the solvent. However, the situation seems rather to be reversed for Tyr66 in MerP. The CD experiments suggested that MerP has a well ordered tertiary structure due to the characteristic spectrum observed in the near-UV CD spectrum. However, this fine structure of the spectrum is completely abolished when 3 M GuHCl is added, signifying that the tertiary structure is to a large extent disrupted under these conditions. The same conclusions could be drawn about the secondary structure from CD measurements in the far-UV spectrum. In the folding experiment it was found that there are no hidden phases before the experimental dead time of 2 ms. However, the observable change in the intrinsic fluorescence upon folding occurs in two exponential phases with apparent rate constants of 1.2 s-1 and 0.053 s-1 respectively. Similar biphasic behaviour was found in the unfolding experiment with apparent rate constants of 2.7 s-1 and 0.012 s-1, respectively. These kinetic observations of MerP were found to be best explained using the following scheme:

which shows that the slow phase occurs due to cis-trans isomerization around a particular X-Pro peptide bond (see Figure 12). In this scheme Uc is the unfolded form in the cis-conformation and Ut is the unfolded form in the trans-conformation.

Figure 12. The two isomeric states of the proline peptide bond.

Cα C N

Cα+1 O

Cα C N

Cα+1

O Trans (62-94 %) Cis (6-38 %)

kt → c

kc → t

UT UC N 0.053

0.012

1.2

1.8

32

In an unfolded population of protein molecules, equilibrium eventually will be reached where the X-Pro peptide bonds of most of the molecules will be in the trans-conformation. However, the bonds of 6-38 % of the molecules will be in the cis-conformation depending on the identity of X.139 MerP contains three proline residues, two of which are located in loop 1, a relatively flexible loop (see Paper II). However, the third proline is located in loop 5, which is a rigid turn structure (see Paper II) and more interestingly this proline is adjacent to Tyr66. Thus, it seems reasonable to conclude that the cis-trans relaxation around Tyr66-Pro67 is responsible for the slow folding/unfolding phases. Cis-trans isomerization is a rather slow process compared to folding/unfolding events. Using pentapeptides it has been found that the rate constants kc→t and kt→c range from 3.2 × 10-3 s-1 to 3.2 × 10-3 s-1 and 4.6 × 10-4 s-1 to 0.4 × 10-4 s-1, respectively, depending on the conditions and the amino acid that is adjacent to the proline.139 The percentage of cis conformers when Tyr is the adjacent amino acid is around 24 %.139

Thus, in the folding process the majority unfolded population (accounting for ≈ 80 % of the total), in which the Tyr66-Pro67 bond is in the trans-conformation, will fold at the higher rate in a single exponential decay. However, the cis-bond in the remaining approximately 20 % of the molecules needs to be converted to trans before they can fold to the native state, thus causing the much slower second phase. This second phase is abolished when the folding process is initiated after only two seconds of unfolding equilibration because cis/trans relaxation will not be reached within such a short equilibrium time. Thus, when jumping back to native conditions all proline peptide bonds will have their native conformation and cis-trans isomerizations will not be rate limiting steps. The same argument holds for the biphasic behaviour seen in the unfolding experiment. In this case all protein molecules start to unfold in a fast single exponential decay. However, equilibration of the unfolded system, in which approximately 20 % of the population adopts a cis-conformation, causes the slower second phase where Ut and Uc clearly have different fluorescence properties. This second phase is abolished when unfolded proteins are refolded for short periods, in a time scale of ms to seconds, before the onset of unfolding, because within the short refolding time only the population in the trans-conformation will gain the native structure and only this population will then be observable in the following unfolding step.

Although data on the folding mechanism of MerP acquired using these conventional methods seem to fit a two-state model perfectly, this does not rule out the possibility that intermediates are formed that are not sufficiently populated to be detectable in these experiments. Therefore, more

33

experiments were needed to acquire a deeper understanding of the folding properties of MerP. II. Native-state HX of MerP at increasing temperatures Introduction In the studies discussed in this paper the folding energy landscape of MerP was further analysed by monitoring native state hydrogen exchange as a function of temperature in the range 7 to 55 °C. The temperature dependence of the hydrogen exchange allowed us to determine ∆G (from equations (3) and (23)) and ∆H and ∆Cp (from equation (13)) at amino acid residue resolution for the conformational processes that permit hydrogen exchange. Interestingly, analysis of the data from these experiments revealed the existence of partially unfolded structures (i.e. intermediates) that might contribute to the unusually slow folding of MerP. Exchange results The results showed that for 11 residues, the free energy (∆GHX) approximates to the free energy of thermal unfolding of MerP as determined by far-UV CD measurements.140 These residues (coloured blue in Figure 13) are mainly located in β-strand 3, the structure around the amide protons of Ala8 and Ser69, the C-terminal end of helix 2 and the beginning of loop 5. Clearly, these parts constitute the most stable region of the structure. Another 15 amide protons exchange within a broad spectrum of free energies that are lower than the free energy for the global unfolding i.e. the exchange at these positions is dependent on a mixed mechanism involving a temperature-dependent unfolding with lower free energy than the global unfolding. The mixed exchange may involve parts of the protein of varying sizes. Thus, the exchange at these positions occurs from partially unfolded structures. These structural elements are mainly found at residues in β-strand 1 that have hydrogen bonds with residues in β-strand 3, and at residues in helices 1 and 2 (coloured green in Figure 13). For 23 amide protons the exchange is mainly dependent on local fluctuations, i.e. the free energy of exchange is quite low and insensitive to temperature (coloured red in Figure 13) or the exchange is too fast to measure at any of the tested temperatures except 7 °C and 15 °C (coloured orange in Figure 13). At 12 other amide protons (coloured yellow in Figure 13) the exchange was too rapid to be observed. Most of the amide protons showing either of these two types of exchange behaviour are located in loops 1-4, the beginning of helices 1 and 2, and -strands 2 and 4 (the two strands at the edge of the -sheet).

34

Figure 13. Exchange mechanisms at individual amide protons of MerP. Amide protons that exchanged by global unfolding are coloured blue, amide protons that show a mixed exchange pattern are coloured green and amide protons where the exchange was relatively insensitive to the temperature are coloured red. Protons where exchange was too rapid to measure at any of the test temperatures except 7 and 15 °C are coloured orange and those that exchanged too rapidly to be observed are coloured yellow. Amide protons for which exchange could not be determined due to spectral overlap are coloured white. Carbon atoms are coloured grey, oxygen atoms are coloured black and the amide nitrogens are coloured light grey. H-bonds in the secondary structure of oxidized MerP, as determined by NMR, are shown by red dotted lines.

35

Based on this exchange pattern it was possible to analyse in detail the stability in different substructures of MerP. Moreover it was also possible to study enthalpies of exchange and to analyze the relationship between ∆Cp and the extent of buried surface. Finally, the study made it possible to speculate about features of the energy landscape of MerP Detailed analysis of secondary structure elements β-sheet: In the analysis of the interactions between β-strands 1 and 4 it was found that although a global unfolding process is needed for complete disruption of the interactions between these strands, the unfolding event seems to start by fraying from the C-terminal end of β-strand 4. Consideration of the exchange pattern for the amino acids in -strand 3 and amino acids in -stands 1 and 2 involved in hydrogen bonding to this strand indicated that β-strand 3 is the most stable part of this β-structure. Thus, the NH-groups in β-stand 3 seem to be substantially more protected by side chains than those in β-stands 1 and 2 after partial unfolding events that separate β-strand 3 from either β-strand 1 or 2.

α-helices: The exchange pattern of helix 1 shows that the stability increases from the N-terminal to the middle part of the helix, around Ala24, and then remains fairly constant towards the C-terminal except at Lys27, which exchanges by local fluctuation (i.e. with rather low energy). The results suggest that the amide bond of Lys27 can be exposed to solvent without a simultaneous exchange competent-exposure of the neighbouring hydrogen bonds. Complete unfolding of helix 1 appears to require global unfolding of the protein, but the helix can also partly unfold by fraying from the N-terminal end, as earlier observed in studies of other proteins.109,141 The stability of the structure of α-helix 2 was found to gradually increase from Val54 to Ala62. Although the amide protons at the N-terminal of helix 2 appear to be able to exchange by local unfolding, complete unfolding of the helix requires global unfolding of the protein. In addition, it was found that the individual stability of the -helices is not correlated to the size of their contact area with the rest of the protein structure. Thus, it seems that more specific, long-range interactions are responsible for the stabilisation of the helices.

Loops: The stability of the five loops in MerP increases in the following order: 3 < 1 < 2 < 4 < 5. While the amino acids in loops 1-4 exchange either by a local unfolding process or by partial unfolding of the protein the exchange pattern for loop 5 shows that a global unfolding event is required for complete exchange-competent exposure of the structure. In addition, the

36

exchange behaviour for Gly65, Tyr66 and Ser69, all of which are associated with loop 5, connecting helix 2 and β-strand 4, indicate that a residual structure comprising much of this area has a high frequency of occurrence in the denatured ensemble. This is because the exchange curves for these residues exceed the global unfolding curve at higher temperatures. Improvement of the secondary structure model From the exchange data and a deeper analysis of the structure using MolMol,142 additional hydrogen bonds were suggested, enabling a proposal refinements of the secondary structure to be made: β-strands 1 and 3 were extended by a hydrogen bond between Val9-NH and Arg 41-CO; helix 1 was extended by one residue at the N-terminal end where Thr20 forms a capping H-bond to its side chain; helix 2 was extended by one residue at both ends, where Lys56 (at the N-cap+3 position) is capped by an H-bond to the βCO of Ser53; loop 2 may constitute a type II β turn; in loop 4, hydrogen bond partners were found to the amide protons of Asp49, Thr50, Lys51 and Ala52, and finally the C-terminal end of α-helix 2 might constitute a 310 helix. Enthalpies of exchange Arrhenius analysis of exchange data detected linear behaviour for amide protons that exchange by local fluctuation, and thus ∆Hop could be calculated from the slopes of Arrhenius plots. The values of ∆Hop for these residues are in the range of 4 – 14 kcal/mol. From measurements of the enthalphy for helix formation it has been proposed that the peptide H-bond in an alanine helix is favourable for folding when it is exposed to solvent (∆H = –0.95 kcal/mol residue).33 However, using an electrostatic approach to evaluate the energetic role of H-bonds it has been found that the electrostatic solvation free energy (ESF) of a central peptide group in a solvent-exposed 15-residue alanine helix is -2.5 kcal/mol,33 which can be considered to be enthalpic.143 Most of this value (around -2.0) kcal/mol is contributed by interactions between water and the peptide CO group.33 In contrast, there might be a desolvation penalty for H-bonded peptide groups that are formed in the interior of the protein because they are buried away from contact with water,144 so buried H-bonds might rather destabilise the protein molecule (∆H = 1.55 kcal/mol residue).33 The ESF values can be larger than -2.5 kcal/mol for H-bonded peptide groups that are more exposed to water than those in the solvent-exposed alanine helix (≤ 9.5 kcal/mol),33 and the peptide groups in all residues that exchange by local fluctuation are partly exposed to the surface. When comparing the ∆Hop values for these amino acids with ESF values it seems that only a few

37

hydrogen bonds have to be opened in local fluctuations prior to exchange, indicating that the structural rearrangements required for exchange to occur at these residues are minor. For amide protons that exchange through a mixed or global mechanism, the enthalpy of the unfolding (∆Hm) is in the range of 38 – 62 kcal/mol. Clearly, therefore, a considerable number of interactions are disrupted in the process preceding exchange of these residues. The Arrhenius plots show the mixed and global exchanges to be markedly non-linear, indicating a shift in the relative contributions from different interactions to the enthalpy at different temperatures. Global and mixed unfolding events involve massive exposure of hydrophobic surface (∆Cp = 470 – 1100 kcal mol-1 K-1) and the enthalpy of hydration of non-polar groups increases markedly with increasing temperature.145 Thus, the observed curvature in the Arrhenius plots for global and mixed unfolding indicates that interactions involving non-polar groups make major enthalpic contributions to global stability. Relationship between ∆Cp and the buried surface A correlation was found between the percentages of buried surface for the residues in α-helices 1 and 2 and their ∆Cp values, i.e. the ∆Cp assigned to each individual residue mainly reflects changes in the buried hydrophobic surface of its own side chain. On the other hand, no correlation between the buried surface of individual residues and ∆Cp in the β-sheet was observed, i.e. the ∆Cp assigned to each individual residue in the β-sheet may also include contributions from changes in buried hydrophobic surface among neighbouring side chains. The observed discrepancy between the correlations between buried surface and ∆Cp for the helices and the β-sheet indicates that the unfolding of the helices is much less cooperative than for the β-sheet. The energy landscape of MerP The major conclusion from the results presented in this paper is that, although earlier folding studies indicated that MerP behaves as a “two-state folder”, partially unfolded forms of MerP are responsible for the observed hydrogen exchange reactions of several peptide NH protons. If these partially unfolded species correspond to intermediates in the folding reaction the energy landscape of MerP might not be smooth (Figure 5a), as could be expected for a two-state folding mechanism with no intermediates, but rather rugged due to the occurrence of partly unfolded structures (Figure 5b). The finding that these partially unfolded species are not connected to a specific structural element with well-defined energy indicates that several rate-limiting chain conformations may exist along different folding trajectories that occur in parallel. Finally, it is tempting to speculate that the

38

slow folding rate of MerP might reflect a complex search process in the unfolding conformational space in which the intermediates represent “traps”, i.e. several intermediate conformations can be visited in the folding process and therefore the folding rate is slow. III. Native-state HX of MerP at increasing concentrations of GuHCl Introduction

In this paper measurements of native-state HX as a function of GuHCl concentration in the range of 0 – 1.0 M were used to characterise the energy landscape of MerP. By analysis of the data using equation (14) it was possible to determine m-values and the free energy of exchange in the absence of denaturant for individual amide protons. The results confirm the earlier observation that existing intermediates of MerP do not contain sub-structures with well-defined energies. Instead, a broad continuum of stabilities was found throughout the protein. However, there seems to be a weak hierarchical coupling between the energetics and structural elements of the protein since the stabilities increase towards three focal points in the structure. Further analysis of the exchange pattern revealed that the denatured form of the protein is less compact in GuHCl than in H2O. Exchange results Results from the measurements showed that 14 of the amide protons (coloured yellow in Figure 14) exchanged before the first spectrum had been recorded indicating that these protons have more or less the same accessibility to the surrounding solvent in the unfolded and native states of the protein. However, a detailed analysis of the exchange pattern showed that for 16 amide protons (coloured red in Figure 14) there is a weak stability dependence on denaturant concentrations, indicating that the dominant mechanism for exchange under native conditions at these positions involves local fluctuations with the exposure of little new surface. Ten of these protons show a stronger GuHCl dependence at higher denaturant concentrations, signifying that under these conditions another unfolding event with exposure of considerable new surface area starts to dominate the exchange mechanism. Yellow and red positions in the structure are mostly found in low stability areas of the protein, i.e. in loops 1-4, the beginning of helices 1 and 2 and β-strands 2 and 4 (see Paper II). For 29 other amide protons a strong denaturant dependence can be observed over the whole range of GuHCl concentrations, indicating that larger portions of the protein structure are involved in the unfolding events preceding exchange, involving extensive exposure of new surface areas. Thirteen of these amide protons (coloured green in Figure 14) show linear

39

dependency according to equation (14). However, the remaining 16 amide protons (coloured blue in Figure 14) show pronounced upward curvatures at lower GuHCl concentrations (< 0.5 M) according to equation (15).

Figure 14. Exchange patterns at individual amide protons of MerP. Colour code: blue, amide protons for which the exchange curve is curved upward; green, amide protons for which exchange is linearly dependent on the denaturant concentration; red, amide protons that exchange during local fluctuations; yellow, amide protons that exchange too rapidly to be observed; white, amide protons for which the exchange cannot be characterised, due to spectral overlap. Carbon atoms are coloured grey, oxygen atoms black and amide nitrogens light grey. H-bonds in the secondary structure of oxidized MerP, as determined by NMR, are shown by red dotted lines.

40

Based on this exchange pattern it was possible to analyse the difference in the compactness of the unfolded ensemble in H2O and GuHCl. In addition it was possible to examine the distribution of partially unfolded forms and to discuss the existence of structural and energetic cores in MerP and their consequences for the shape of the folding energy landscape of MerP. Residual structure of the unfolded ensemble in water

Detailed analysis of the upward curvature exchange pattern indicates that in the H2O-denaturated state the part of the structure coloured blue in Figure 3, Paper III, is up to five times more resistant to exchange, and thus more protected, than the same region in the GuHCl-denatured state. In other words, the unfolded polypeptide in water has a more compact structure than the unfolded polypeptide in GuHCl. This is consistent with a hypothesis that the upward curvature results from an overestimation of the krc value in experiments at low denaturant concentrations. This curvature can be corrected by reducing the krc value by a factor of up to five, indicating that these amino acids are up to five times more protected from exchange in the unfolded polypeptide than in an unstructured polypeptide. This feature is found in the most central parts of the secondary structures. In other parts of the structure there is no difference in the compactness between the GuHCl- and H2O- denatured states. Distribution of partially unfolded forms with different energies For 38 of the amide protons it was possible to determine the free energy value at 0 M GuHCl (∆G(u,0)) and the coupled m-value for the GuHCl-dependent unfolding process. The stabilities for the individual residues and their m-values were found to range between 2.6 to 4.7 kcal/mol and 1.3 to ~ 3.5 kcal mol-1 M-1, respectively. These broad distributions of denaturant-dependent ∆G(u,0) energies and m-values in MerP are consistent with the occurrence of a large number of energy levels between the native and unfolded forms, indicating the existence of partially unfolded forms. It has been proposed that coincidence of the ∆G(u,0) and m values for neighbouring residues into distinct isotherms implies a subdivision of the protein into different PUFs (i.e. partially unfolded forms that contain local substructures with well-defined energies).99,100 In these systems (especially for cyt c) it has been shown that PUFs identified using native-state hydrogen exchange resemble kinetic intermediates and occur sequentially in a stepwise manner along the folding pathway,20,100,114 a model that is consistent with the “classical view” of protein folding.

The observation that residues with similar energies do not cluster in well-defined areas in MerP shows that it does not unfold via the formation of a

41

few discrete partially unfolded forms. Instead, there seem to be large numbers of essentially unrelated (uncoupled) partially unfolded forms with similar energies. This implies that the energy landscape of MerP may be rugged and many structurally dissimilar states with similar energies may occur on different folding trajectories that are more or less uncoupled from each other. The slow folding rate of MerP might be coupled to the extensive visitation of large numbers of partly unrelated conformations during the folding process. Structural and energetic core of MerP The range of observed ∆GHX (H2O) values is rather broad, varying from 0.6 to 4.9 kcal/mol. A search for correlations between ∆GHX values and distances between residues in the structure resulted in apparently linear correlations between: (i) the ∆GHX value and the distance to the Cα of Leu25 for the residues in helix 1, (ii) the ∆GHX value and the distance to the Cα of Asp63 for the residues in helix 2, and (iii) the ∆GHX value and the distance to the Cα of Leu7 for the most stable part of the β-sheet. Thus, there are weak indications that there are at least three focal points (residues 7, 25 and 63) that are close to centres to which additional structural elements can be gradually added. However it remains to be investigated if these observed centres of stability are coupled in any way to nucleation centres in the kinetic folding process. Conclusions

Fascinatingly, the data from this study are clearly consistent with earlier conclusions drawn from the temperature-dependent hydrogen exchange study (see Paper II). The energy landscape of MerP is most probably rugged with the existence of different folding trajectories that occur in parallel (as suggested by the detection of unrelated partially unfolded forms in both studies). However, the finding of three centres in the protein that might be important for the formation of native structure during the folding process indicates that the different folding trajectories might merge into a few major pathways that lead to the native state. The search for these routes might be complex, slowing down the folding process.

Finally the hydrogen exchange pattern in MerP did also reveal that the most central parts of the secondary structures have a more compact structure when the protein is unfolded in H2O than when the protein is unfolded in GuHCl, i.e. the central core of the unfolded protein is more protected from exchange in an H2O environment than in a GuHCl environment.

42

IV. Multivariate analysis of 2D NMR data Introduction The spectral shifts in an NMR spectrum are very sensitive to the molecular environment of specific nuclei in the molecule being examined. Thus, spectral changes for individual amide protons that occur when a protein unfolds can provide information at amino acid residue resolution about early structural changes during the unfolding process. However, a general problem when analysing NMR spectra that reflect variations in the environment of target molecules is that different resonances are affected to differing extents. Often a few resonances that display the largest frequency changes are selected as probes to reflect the examined variation. This paper describes the use of multivariate analysis (MVA) to analyse multiple NMR spectra recorded under various folding conditions for MerP and also how the results can be interpreted in terms of structural changes in the protein molecule. Method description The data used in the analysis consist of 11 1H-15N ge-HSQC spectra, recorded at varying GuHCl concentrations. By increasing the denaturant concentration stepwise from 0.35 to 2.9 M, the transition from predominantly native to unfolded protein was induced. The collected 2D-NMR time-domain data were then subjected to multivariate analysis. In a first step, PCA calculations were performed on the 2D-NMR time domain data in order to quantify how similar the different spectra are to one another and thus obtain a statistical representation of the spectral changes. Interestingly, the first three principal components of the PCA analysis revealed the occurrence of at least three main changes, i.e. the addition of denaturant induces spectral changes which do not correspond to the main two-state unfolding process of the protein found in Paper I. Thus, the 11 NMR measurements were divided into three sub-groups covering the following denaturant concentrations ranges; 0.4-1.4 M, 1.6-2.1 M and 2.1-2.9 M GuHCl. In a second step the 2D-NMR time domain data of each subgroup were subjected to a PLS analysis. In this step the GuHCl concentration was used as the response, in order to detect spectral changes related to the denaturant concentration. To visualise the spectral variation in a single NMR spectrum a Fourier transformation of the first PLS weight vector for each sub-group was obtained, resulting in three 2D-NMR spectra, hereafter referred to as PLS-spectra. These three PLS-spectra only reflect differences between the various recorded NMR spectra (within each sub-group) due to conformational changes induced by the denaturant since resonances that

43

were not affected by the addition of denaturant would be ignored in this procedure. Results and discussion

The main feature of the PLS-spectra is that they all resemble a native NMR spectrum and the visible resonances from the amide protons either appear as positive (blue), if their intensities are positively correlated to the GuHCl concentration, or negative (red), if their intensities are negatively correlated to the GuHCl concentration. In the PLS-spectrum of sub-group 1 (0.4-1.4 M GuHCl) it can be seen that the changes in the spectra for the native resonances are predominantly represented by frequency shifts. This is manifested by the peaks having both negative (red) and positive (blue) contributions. Such behaviour is expected for nuclei that are involved in conformational exchanges that are rapid on the NMR time-scale i.e. within a few ms. The rapid exchange indicates that the populated conformations are easily accessible, and thus do not represent the major unfolding process. Presumably, the bulk of these spectral changes are due to the already known affects of GuHCl on resonance frequencies. However, in contrast to the general behaviour of increases in proton chemical shifts, the frequency shifts for the amide protons of Phe38 and Glu39 decrease upon addition of denaturant. Therefore, these two residues can adopt a non-native conformation prior to the main denaturation of the protein. Phe38 and Glu39 are located in loop 3, which has previously been reported to adopt different conformations in the oxidized and reduced forms of MerP,129,146 implying that this loop can adopt different conformations of similar energy. Thus, the anomalous spectral changes for these two residues might reflect local inter-conversion between these conformations. Further analysis of the PLS-spectrum of sub-group 1 revealed that in contrast to the general invariance in the 15N frequency, the resonances for some amide protons display significant decreases in their 15N frequencies following addition of denaturant. Interestingly, most of these amino acids are located in β-strands 2 & 4, and are hydrogen bonded to β-strands 3 & 1, respectively. The 15N frequency changes suggest that there are native-like conformations where the patterns of the hydrogen bonds between β-strands 1 & 4 and 2 & 3 differ from their patterns in the native state of the protein. Interestingly, such frequency changes are not observed for the amide protons connecting β-strands 1 and 3. This behaviour is consistent with observations from the temperature-dependent hydrogen exchange study suggesting that β-strand 3 is part of the most stable region of the protein structure (see Figure 13). The main features of the PLS-spectrum of subgroup 2 (1.6-2.1 M GuHCl) is that large numbers of purely negative peaks represent the native

44

state, while unfolded states of MerP appear as less well-resolved positive peaks. The appearance of purely red and blue peaks suggests that the resonances reflect processes that are involved in slow exchange on the NMR time-scale due to significant energy barriers between the conformational states involved. Thus, these peaks represent inter-conversions between native and unfolded states of the protein, and this GuHCl range corresponds to the main transition from the native to the unfolded protein. However, the integrals of the negative peaks are substantially lower for eight amino acids found in loops 1 and 3 and at the N-terminal part of helix 1, indicating that this structural region of the protein has adopted, to a certain extent, a non-native conformation prior to this transition. Further support for this conclusion comes from the observation that these amino acids, along with Ala16, are absent in the PLS-spectrum of sub-group 3 (2.1-2.9 M GuHCl). Interestingly, in Paper II it was shown that these substructures are involved in the most flexible regions of the protein molecule (see Figure 13). In contrast, several amino acids can still be observed in the PLS-spectrum of sub-group 3, most of which are located in the secondary structure, indicating that the secondary structure of MerP is not completely ruptured, even at the higher denaturant concentrations used in this study. Conclusions

The fluorescence and CD equilibrium unfolding analyses of MerP revealed only one transition, between approximately 0.7-2.5 M GuHCl (see Paper I). However, PCA analysis of 2D NMR data acquired in the same region allowed this transition to be divided into three different sub-groups, and indicated that the main structural disruption occurs between 1.6 and 2.1 M GuHCl. In addition, using PLS analysis on each sub-group it was possible to detect structural changes at distinct regions in the protein. Thus, it has been shown that multivariate data analysis can be successfully applied to 2D-NMR time domain data revealing structural changes at amino acid resolution. Comparison of proteins with similar topology to MerP

Earlier studies by Plaxco et al proposed that the contact order (CO), folding rate (kf) and transition state placement (θm) of single domain proteins are correlated,61 i.e. that proteins with mainly local interactions (low contact order) tend to fold more rapidly and to have less compact folding transition states (lower θm values) than those with more non-local interactions. Based on this study the cited authors proposed a model of

45

protein folding in which the size and stability of the organized polypeptide segments in the transition state are related to the topological complexity of the native state. However, the above study was based on a wide range of single domain proteins with different topologies. Table 1 shows the stability, kinetics and contact order for MerP and five other two-state proteins that have identical or very similar topologies. These values are compared with each other to see if the model suggested by Plaxco et al is valid within this homologous protein group. Table 1. Length k f (s

-1) ∆∆∆∆Gu (kcal/mol)

θθθθmc CO % Ref

MerP 72 1.8 3.5 0.35d 18.9 130,140,147

U1A 102 339 9.9 16.91 148

S6 101 398 8.2 0.70 18.9 130

ADA2h 80 758 4.1 0.74 16.96 137

AcPa 98 0.24 4.9 0.77 21.2 134

HPrb 85 14.9 4.7 0.64 18.35 135 a The β-sheet of this protein contains five strands. b The β-sheet of this protein contains three strands. c θm can be calculated from the ratio of the denaturant dependencies of the free energy of

folding ( eqm from equation (14)) and folding transition state (‡

fm )61 and indicates the

amount of solvent-accessible surface buried in the native state that is also buried in the transition state. Thus, if θm = 0 the transition state is highly unstructured while if θm ≈ 1 the transition state has a very native-like structure. d Calculated from the dependence of the refolding and unfolding rates of MerP on GuHCl147

In a study by Chiti et al134 a negative correlation between lnkf and CO was found for U1A, ADA2h, HPr, MerP and AcP (R2 = 0.96) supporting Plaxco´s model. However when S6 is included in this plot the correlation coefficient drops to 0.70. Interestingly, MerP and S6 have the same contact order, but S6 folds with a 200 times faster rate constant than MerP. Therefore, factors other than the contact order of the structure must play a major role in the observed differences between the folding rates for these two proteins.

The CO and θm data show no association between these two parameters within this subgroup (R2 = 0.01), in contrast to the general pattern found by Plaxco et al.61 This indicates that within this group the average distance in sequence between interacting residues in the native state does not have a significant impact on the compactness of the transition state.

46

From a comparison of a these proteins131,133,134,149 it seems that the folding nuclei within this group are quite diverse and can occur in more or less any part of the protein structure. This diversity may reflect the high diversity of the kinetics and stability of these proteins. Thus, no general rules developed as yet seem to be applicable to the folding properties of proteins with this topology. Conclusions CD and fluorescence analyses of the unfolding/refolding process indicate that MerP behaves as a typical “two-state folder”. However, native-state hydrogen exchange analyses revealed the existence of partially unfolded forms, which indicates that the energy landscape of MerP is rugged, with the existence of several folding trajectories that occur in parallel. The different folding routes might merge into three major routes leading to the native state. The extensive search in the folding landscape might be the reason for the rather slow folding rate of MerP.

Hydrogen exchange in MerP did also reveal that the unfolded protein contains a central core which is more compact in an H2O environment than in a GuHCl environment. Multivariate data analysis to detect conformational changes induced by GuHCl indicated that the native structure of a discrete region involved in the most flexible part of the structure, i.e. loops 1 and 3 and the N-terminal part of helix 1 (Paper II), is ruptured at relatively low denaturant concentrations (< 2.1 M GuHCl) while the native structure of the secondary elements is still not completely ruptured at 2.9 M GuHCl. Finally, by comparing the stability, kinetics, contact order and folding nuclei for six proteins (including MerP) with similar topology it was found that the folding properties for this topology are quite diverse and no general rule can currently be applied to predict their folding behaviour.

47

Acknowledgements Jag önskar att uttrycka min stora tacksamhet till alla som har hjälpt och stöttat mig under mina doktorandstudier. Särskilt vill jag tacka följande: • Min handledare Nalle för allt stöd under dessa år, för alla vetenskapliga

och sociala diskussioner och för din och Lenas stora gästvänlighet (jag hoppas att jag inte varit alltför svår att uppfostra).

• Min handledare Ingmar för många trevliga och mycket givande diskussioner, tack för ditt stöd och för trevlig samvaro i Västanbäck.

• Anders Ö för ditt otroligt stora tålamod och all hjälp, jag kan inte uttrycka det i ord.

• Nuvarande och gamla medlemmar i min forskargrupp. Tack Per, Aron och Annika för all hjälp under mina första trevande år på labb och vid datorn samt för trevligt umgänge. Tack även till Martin för all forskarhjälp och trevliga pratstunder (det finns ett liv även efter detta!).

• Mikael Oliveberg för all hjälp med mina ”veckningsproblem”. • Christina för alla ”pep-talks”, delad glädje är dubbel glädje. • Katarina , Viola och Anna-Märta för all hjälp med både det ena och det

andra samt för mycket trevliga stunder kring kaffebordet, jag kommer att sakna er!

• Katja for your never ending optimism and your inspiration. • Alvaro for just being you! • Alla övriga personer på Biokemi (nuvarande och gamla) för trevlig

gemenskap både i fikarummet och i forskarsammanhang. • Ludmilla , Vladimir and Mantas for interesting fibril-studies of MerP

and nice company. • Alla medlemmar i Yogasällskapet för roliga och givande träningspass

som får mig att hålla fötterna upp och huvudet ner. • Janik Hayrapetyan for sharing the gift of yoga. I miss you! • Svärmor Kerstin för all hjälp med barnen. Utan dig hade detta inte varit

möjligt (ett tack även till svärfar Egil). • Ingalill och Stig för trevliga grillkvällar och övrig samvaro. Norge nästa! • Marie och Anncha för trogen vänskap i vått och tort. • Alla vänner i Vännäs missionsförsamling, ingen nämnd ingen glömd.

Tack för att ni finns! • Mamma och Pappa för att ni alltid ställer upp för mig. Andreas och

Angelica för att ni inte överger mig trots bortglömda födelsedagar och uteblivna telefonsamtal. Peter för ditt tålamod, all hjälp och för att du finns vid min sida. William och Ramona för att jag fått äran att vara er mamma. Jag älskar er allihop!

48

References 1. Pauling, L., Corey, R. B. & Branson, H. R. (1951). The structure of

proteins: two hydrogen-bonded helical configurations of the polypeptide chain. Proc. Natl. Acad. Sci. USA, 37, 205-211.

2. Pauling, L. & Corey, R. B. (1951). Configurations of polypeptide chains with favored orientations around single bonds: two new pleated sheets. Proc. Natl. Acad. Sci. USA, 37, 729-740.

3. Sanger, F., Thompson, E. O. & Kitai, R. (1955). The amide groups of insulin. Biochem. J. 59, 509-518.

4. Kendrew, J. C., Bodo, G., Dintzis, H. M., Parrish, R. G., Wyckoff, H. & Phillips, D. C. (1958). A three-dimensional model of the myoglobin molecule obtained by x-ray analysis. Nature, 181, 662-666.

5. Perutz, M. F., Rossman, M. G., Cullis, A. F., Muirhead, H., Will, G. & North, A. C. T. (1960). Structure of haemoglobin. A three-dimensional Fourier synthesis at 5.5Å resolution, obtained by X-ray analysis. Nature, 185, 416-422.

6. Richards, F. M. (1991). The Protein Folding Problem. Scientific Am. January, 34-41.

7. Kim, P. S. & Baldwin, R. L. (1982). Specific intermediates in the folding reactions of small proteins and the mechanism of protein folding. Annu. Rev. Biochem. 51:459-89, 459-489.

8. Ohgushi, M. & Wada, A. (1983). 'Molten-globule state': a compact form of globular proteins with mobile side-chains. FEBS Lett. 164, 21-24.

9. Wu, L. C., Peng, Z. Y. & Kim, P. S. (1995). Bipartite structure of the alpha-lactalbumin molten globule. Nat. Struct. Biol. 2, 281-286.

10. Dobson, C. M. (1991). Characterization of protein folding intermediates. Curr. Opin. Struct. Biol. 1, 22-27.

11. Dobson, C. M. (1995). Finding the right fold. Nat. Struct. Biol. 2, 513-517.

12. Dobson, C. M., Evans, P. A. & Radford, S. E. (1994). Understanding how proteins fold: The lysozyme story so far. Trends Biochem. Sci. 19, 31-37.

13. Fersht, A. R. (1995). Optimization of rates of protein folding: The nucleation- condensation mechanism and its implications. Proc. Natl. Acad. Sci. USA, 92, 10869-10873.

14. Matthews, C. R. (1993). Pathways of protein folding. Annu. Rev. Biochem. 62, 653-683.

15. Ptitsyn, O. B. (1994). Kinetic and equilibrium intermediates in protein folding. Protein Eng. 7, 593-596.

49

16. Roder, H. & Colón, W. (1997). Kinetic role of early intermediates in protein folding. Curr. Opin. Struct. Biol. 7, 15-28.

17. Kim, P. S. & Baldwin, R. L. (1990). intermediates in the folding reactions of small proteins. Annu. Rev. Biochem. 59, 631-660.

18. Creighton, T. E., Darby, N. J. & Kemmink, J. (1996). The roles of partly folded intermediates in protein folding. FASEB J. 10, 110-118.

19. Clarke, A. R. & Waltho, J. P. (1997). Protein folding pathways and intermediates. Curr. Opin. Biotechnol. 8, 400-410.

20. Bai, Y. (2003). Hidden intermediates and levinthal paradox in the folding of small proteins. Biochem. Biophys. Res. Commun. 305, 785-788.

21. Krantz, B. A., Mayne, L., Rumbley, J., Englander, S. W. & Sosnick, T. R. (2002). Fast and slow intermediate accumulation and the initial barrier mechanism in protein folding. J. Mol. Biol. 324, 359-371.

22. Sanchez, I. E. & Kiefhaber, T. (2003). Evidence for sequential barriers and obligatory intermediates in apparent two-state protein folding. J. Mol. Biol. 325, 367-376.

23. Miller, E. J., Fischer, K. F. & Marqusee, S. (2002). Experimental evaluation of topological parameters determining protein-folding rates. Proc. Natl. Acad. Sci. U S A, 99, 10359-10363.

24. Plaxco, K. W., Simons, K. T., Ruczinski, I. & Baker, D. (2000). Topology, stability, sequence, and length: defining the determinants of two-state protein folding kinetics. Biochemistry, 39, 11177-11183.

25. Makarov, D. E. & Plaxco, K. W. (2003). The topomer search model: A simple, quantitative theory of two-state protein folding kinetics. Protein Sci. 12, 17-26.

26. Makarov, D. E., Keller, C. A., Plaxco, K. W. & Metiu, H. (2002). How the folding rate constant of simple, single-domain proteins depends on the number of native contacts. Proc. Natl. Acad. Sci. U S A, 99, 3535-3539.

27. Bachmann, A. & Kiefhaber, T. (2001). Apparent two-state tendamistat folding is a sequential process along a defined route. J. Mol. Biol. 306, 375-386.

28. Dill, K. A. & Fersht, A. R. (1996). Folding and binding - Editorial overview. Curr. Opin. Struct. Biol. 6, 1-2.

29. Dobson, C. M. (2003). Protein folding and misfolding. Nature, 426, 884-890.

30. Dobson, C. M. (1999). Protein misfolding, evolution and disease. Trends Biochem. Sci. 24, 329-332.

31. Murphy, K. P. (1995). Noncovalent forces important to the conformational stability of protein structures. In Methods in

50

Molecular Biology, Protein stability and folding (Shirley, B. A., ed.), pp. 1-34, Humana Press, New Jersey.

32. Jeffrey, G. A. & Saenger, W. (1991). Hydrogen bonding in biological structures. Springer-Verlag, Berlin.

33. Baldwin, R. L. (2003). In search of the energetic role of peptide hydrogen bonds. J. Biol. Chem. 278, 17581-17588.

34. Dill, K. A. (1990). Dominant forces in protein folding. Biochemistry, 29, 7133-7155.

35. Harpaz, Y., Gerstein, M. & Chothia, C. (1994). Volume changes on protein folding. Structure, 2, 641-649.

36. Shortle, D. (1992). Mutational studies of protein structures and their stabilities. Q. Rev. Biophys. 25, 205-250.

37. Wallon, G., Kryger, G., Lovett, S. T., Oshima, T., Ringe, D. & Petsko, G. A. (1997). Crystal structures of Escherichia coli and Salmonella typhimurium 3-isopropylmalate dehydrogenase and comparison with their thermophilic counterpart from Thermus thermophilus. J. Mol. Biol. 266, 1016-1031.

38. Stigter, D., Alonso, D. O. & Dill, K. A. (1991). Protein stability: electrostatics and compact denatured states. Proc. Natl. Acad. Sci. U S A, 88, 4176-4180.

39. Powlowski, J. & Sahlman, L. (1999). Reactivity of the two essential cysteine residues of the periplasmic mercuric ion-binding protein, MerP. J. Biol. Chem. 274, 33320-33326.

40. Spolar, R. S., Livingstone, J. R. & Record, M. T., Jr. (1992). Use of liquid hydrocarbon and amide transfer data to estimate contributions to thermodynamic functions of protein folding from the removal of nonpolar and polar surface from water. Biochemistry, 31, 3947-3955.

41. Robertson, A. D. & Murphy, K. P. (1997). Protein Structure and the Energetics of Protein Stability. Chem. Rev. 97, 1251-1268.

42. Pace, C. N., Shirley, B. A., McNutt, M. & Gajiwala, K. (1996). Forces contributing to the conformational stability of proteins. FASEB J. 10, 75-83.

43. Fersht, A. R. (1999). Protein Stability. In Structure and mechanism in protein science, Aguide to enzyme catalysis and protein folding. pp.508-536, Freeman, New York.

44. Dill, K. A. & Chan, H. S. (1997). From Levinthal to pathways to funnels. Nat. Struct. Biol. 4, 10-19.

45. Levinthal, C. (1969). How to fold graciously. pp. 22-24, University of Illinois Press, Urbana, Illinois.

51

46. Konermann, L. & Simmons, D. A. (2003). Protein-folding kinetics and mechanisms studied by pulse-labeling and mass spectrometry. Mass Spectrom. Rev. 22, 1-26.

47. Chan, H. S. & Dill, K. A. (1998). Protein folding in the landscape perspective: chevron plots and non-Arrhenius kinetics. Proteins, 30, 2-33.

48. Miller, D. W. & Dill, K. A. (1995). A statistical mechanical model for hydrogen exchange in globular proteins. Prot. Sci. 4, 1860-1873.

49. Arai, M. & Kuwajima, K. (2000). Role of the molten globule state in protein folding. Adv. Protein. Chem. 53, 209-282.

50. Kamagata, K., Arai, M. & Kuwajima, K. (2004). Unification of the Folding Mechanisms of Non-two-state and Two-state Proteins. J. Mol. Biol, 339, 951-965.

51. Ptitsyn, O. B. (1995). Molten globule and protein folding. Adv. Protein Chem. 47, 83-229.

52. Jackson, S. E. (1998). How do small single-domain proteins fold? Folding & Design, 3, R81-R91.

53. Schonbrun, J. & Dill, K. A. (2003). Fast protein folding kinetics. Proc. Natl. Acad. Sci. U S A, 100, 12678-12682.

54. Ozkan, S. B., Dill, K. A. & Bahar, I. (2002). Fast-folding protein kinetics, hidden intermediates, and the sequential stabilization model. Protein Sci. 11, 1958-1970.

55. Wetlaufer, D. B. (1973). Nucleation, rapid folding, and globular intrachain regions in proteins. Proc. Natl. Acad. Sci. U S A, 70, 697-701.

56. Ptitsyn, O. B. & Rashin, A. A. (1975). A model of myoglobin self-organization. Biophys Chem. 3, 1-20.

57. Karplus, M. & Weaver, D. L. (1976). Protein-folding dynamics. Nature, 260, 404-406.

58. Dill, K. A. (1985). Theory for the folding and stability of globular proteins. Biochemistry, 24, 1501-1509.

59. Fersht, A. R. (1997). Nucleation mechanisms in protein folding. Curr. Opin. Struct. Biol. 7, 3-9.

60. Neira, J. L. & Fersht, A. R. (1999). Exploring the folding funnel of a polypeptide chain by biophysical studies on protein fragments. J. Mol. Biol. 285, 1309-1333.

61. Plaxco, K. W., Simons, K. T. & Baker, D. (1998). Contact order, transition state placement and the refolding rates of single domain proteins. J. Mol. Biol. 277, 985-994.

62. Nolting, B. & Andert, K. (2000). Mechanism of protein folding. Proteins, 41, 288-298.

52

63. Daggett, V. & Fersht, A. R. (2003). Is there a unifying mechanism for protein folding? Trends Biochem. Sci. 28, 18-25.

64. Schindler, T. & Schmid, F. X. (1996). Thermodynamic properties of an extremely rapid protein folding reaction. Biochemistry,35, 16833-16842.

65. Privalov, P. L. (1992). Physical basis of the Stability of the folded conformations of proteins. In Protein folding (Creighton, T. E., ed.), pp. 83-126, Freeman, New York.

66. Privalov, P. L. & Khechinashvili, N. N. (1974). A thermodynamic approach to the problem of stabilization of globular protein structure: a calorimetric study. J. Mol. Biol. 86, 665-684.

67. Pace, C. N. (1975). The stability of globular proteins. CRC Crit Rev. Biochem. 3, 1-43.

68. Tanford, C. (1970). Protein denaturation. C. Theoretical models for the mechanism of denaturation. Adv. Protein Chem. 24, 1-95.

69. Pace, C. N., Shirley, B. A. & Thomson, J. A. (1990). Measuring the conformational stability of a protein. In Protein structure, a practical approach (Creighton, T. E., ed.), pp. 311-330, IRL Press, Oxford.

70. Aune, K. C. & Tanford, C. (1969). Thermodynamics of the Denaturation of Lysozyme by Guanidine Hydrochloride. II. Dependence on Denaturant Concentration at 25 grader. Biochemistry, 8, 4586-4590.

71. Nozaki, Y. & Tanford, C. (1970). The Solubility of Amino Acids, Diglycine, and Triglycine in Aqueous Guanidine Hydrochloride Solutions. J. Biol. Chem. 245, 1648-1652.

72. Kirschner, D. A., Abraham, C. & Selkoe, D. J. (1986). X-ray diffraction from intraneuronal paired helical filaments and extraneuronal amyloid fibers in Alzheimer disease indicates cross-beta conformation. Proc. Natl. Acad. Sci. U S A, 83, 503-507.

73. Jimenez, J. L., Guijarro, J. I., Orlova, E., Zurdo, J., Dobson, C. M., Sunde, M. & Saibil, H. R. (1999). Cryo-electron microscopy structure of an SH3 amyloid fibril and model of the molecular packing. EMBO J. 18, 815-821.

74. Kheterpal, I., Zhou, S., Cook, K. D. & Wetzel, R. (2000). Abeta amyloid fibrils possess a core structure highly resistant to hydrogen exchange. Proc. Natl. Acad. Sci. U S A, 97, 13597-13601.

75. Tycko, R. (2001). Biomolecular solid state NMR: advances in structural methodology and applications to peptide and protein fibrils. Annu. Rev. Phys. Chem. 52, 575-606.

53

76. Guijarro, J. I., Sunde, M., Jones, J. A., Campbell, I. D. & Dobson, C. M. (1998). Amyloid fibril formation by an SH3 domain. Proc. Natl. Acad. Sci. U S A, 95, 4224-4228.

77. Litvinovich, S. V., Brew, S. A., Aota, S., Akiyama, S. K., Haudenschild, C. & Ingham, K. C. (1998). Formation of amyloid-like fibrils by self-association of a partially unfolded fibronectin type III module. J. Mol. Biol. 280, 245-258.

78. Chiti, F., Webster, P., Taddei, N., Clark, A., Stefani, M., Ramponi, G. & Dobson, C. M. (1999). Designing conditions for in vitro formation of amyloid protofilaments and fibrils. Proc. Natl. Acad. Sci. U S A, 96, 3590-3594.

79. Caughey, B. & Lansbury, P. T., Jr. (2003). Protofibrils, Pores, Fibrils, and Neurodegeneration: Separating the Responsible Protein Aggregates from the Innocent Bystanders. Annu. Rev. Neurosci. 26, 267-298.

80. Lansbury, P. T., Jr. (1999). Evolution of amyloid: what normal protein folding may tell us about fibrillogenesis and disease. Proc. Natl. Acad. Sci. U S A, 96, 3342-3344.

81. Malisauskas, M., Zamotin, V., Jass, J., Noppe, W., Dobson, C. M. & Morozova-Roche, L. A. (2003). Amyloid protofilaments from the calcium-binding protein equine lysozyme: formation of ring and linear structures depends on pH and metal ion concentration. J. Mol. Biol. 330, 879-890.

82. Goldberg, M. S. & Lansbury, P. T., Jr. (2000). Is there a cause-and-effect relationship between alpha-synuclein fibrillization and Parkinson's disease? Nat. Cell Biol. 2, E115-E119.

83. Haass, C. & Steiner, H. (2001). Protofibrils, the unifying toxic molecule of neurodegenerative disorders? Nat Neurosci. 4, 859-860.

84. Friebolin, H. (1993). Basic one- and two-dimensional NMR spectroscopy. VCH Verlagsgesellschaft, New York.

85. Wuthrich, K. (1986). NMR of proteins and nucleic acids. Wiley Interscience, USA.

86. Zhang, O., Kay, L. E., Olivier, J. P. & Forman-Kay, J. D. (1994). Backbone 1H and 15N resonance assignments of the N-terminal SH3 domain of drk in folded and unfolded states using enhanced-sensitivity pulsed field gradient NMR techniques. J. Biomol. NMR. 4, 845-858.

87. Englander, S. W., Mayne, L., Bai, Y. & Sosnick, T. R. (1997). Hydrogen exchange: The modern legacy of Linderstrom-Lang. Protein Sci. 6, 1101-1109.

54

88. Englander, S. W. (2000). Protein folding intermediates and pathways studied by hydrogen exchange. Annu. Rev. Biophys. Biomol. Struct. 29, 213-238.

89. Bai, Y., Milne, J. S., Mayne, L. & Englander, S. W. (1994). Protein stability parameters measured by hydrogen exchange. Proteins, 20, 4-14.

90. Englander, S. W. & Kallenbach, N. R. (1984). Hydrogen exchange and structural dynamics of proteins and nucleic acids. Quarterly Review of Biophysics, 16, 521-655.

91. Arrington, C. B. & Robertson, A. D. (2000). Kinetics and thermodynamics of conformational equilibria in native proteins by hydrogen exchange. Methods Enzymol. 323, 104-124.

92. Bai, Y., Milne, J. S., Mayne, L. & Englander, S. W. (1993). Primary structure effects on peptide group hydrogen exchange. Proteins, 17, 75-86.

93. Roder, H., Elöve, G. A. & Ramachandra Shastry, M. C. (2000). Early stages of protein folding. In Mechanism of protein folding (Pain, R.H., ed.), pp. 65-95, Oxford university press, New York.

94. Hvidt, A. (1964). A discussion of the ph dependence of the hydrogen-deuterium exchange of proteins. C. R. Trav. Lab Carlsberg. 34, 299-317.

95. Ferraro, D. M. & Robertson, A. D. (2004). EX1 hydrogen exchange and protein folding. Biochemistry, 43, 587-594.

96. Englander, S. W., Sosnick, T. R., Englander, J. J. & Mayne, L. (1996). Mechanisms and uses of hydrogen exchange. Curr. Opin. Struct. Biol. 6, 18-23.

97. Roder, H., Wagner, G. & Wuthrich, K. (1985). Amide proton exchange in proteins by EX1 kinetics: studies of the basic pancreatic trypsin inhibitor at variable p2H and temperature. Biochemistry, 24, 7396-7407.

98. Englander, S. W. (1998). Native-state HX. Trends Biochem. Sci. 23, 378-381.

99. Chamberlain, A. K., Handel, T. M. & Marqusee, S. (1996). Detection of rare partially folded molecules in equilibrium with the native conformation of RNaseH. Nat. Struct. Biol. 3, 782-787.

100. Bai, Y., Sosnick, T. R., Mayne, L. & Englander, S. W. (1995). Protein folding intermediates: Native-state hydrogen exchange. Science, 269, 192-197.

101. Clarke, J. & Fersht, A. R. (1996). An evaluation of the use of hydrogen exchange at equilibrium to probe intermediates on the protein folding pathway. Fold. Des. 1, 243-254.

55

102. Mayo, S. L. & Baldwin, R. L. (1993). Guanidinium chloride induction of partial unfolding in amide proton exchange in RNase A. Science, 262, 873-876.

103. Mullins, L. S., Pace, C. N. & Raushel, F. M. (1997). Conformational stability of ribonuclease T1 determined by hydrogen-deuterium exchange. FASEB J. 11, 322-

104. Yi, Q., Scalley, M. L., Simons, K. T., Gladwin, S. T. & Baker, D. (1997). Characterization of the free energy spectrum of peptostreptococcal protein L. Folding & Design 2, 271-280.

105. Itzhaki, L. S., Neira, J. L. & Fersht, A. R. (1997). Hydrogen exchange in chymotrypsin inhibitor 2 probed by denaturants and temperature. J. Mol. Biol. 270, 89-98.

106. Alexandrescu, A. T., Jaravine, V. A., Dames, S. A. & Lamour, F. P. (1999). NMR hydrogen exchange of the OB-fold protein LysN as a function of denaturant: the most conserved elements of structure are the most stable to unfolding. J. Mol. Biol. 289, 1041-1054.

107. Llinas, M., Gillespie, B., Dahlquist, F. W. & Marqusee, S. (1999). The energetics of T4 lysozyme reveal a hierarchy of conformations. Nat. Struct. Biol. 6, 1072-1078.

108. Chu, R., Pei, W., Takei, J. & Bai, Y. (2002). Relationship between the native-state hydrogen exchange and folding pathways of a four-helix bundle protein. Biochemistry, 41, 7998-8003.

109. Neira, J. L. & Mateu, M. G. (2001). Hydrogen exchange of the tetramerization domain of the human tumour suppressor p53 probed by denaturants and temperature. Eur. J. Biochem. 268, 4868-4877.

110. Chi, Y. H., Kumar, T. K., Chiu, I. M. & Yu, C. (2002). Identification of rare partially unfolded states in equilibrium with the native conformation in an all beta-barrel protein. J. Biol. Chem. 277, 34941-34948.

111. Fuentes, E. J. & Wand, A. J. (1998). Local dynamics and stability of apocytochrome b562 examined by hydrogen exchange. Biochemistry, 37, 3687-3698.

112. Bhuyan, A. K. & Udgaonkar, J. B. (1998). Two structural subdomains of barstar detected by rapid mixing NMR measurement of amide hydrogen exchange. Proteins, 30, 295-308.

113. Parker, M. J. & Marqusee, S. (2001). A kinetic folding intermediate probed by native state hydrogen exchange. J. Mol. Biol. 305, 593-602.

114. Hoang, L., Bedard, S., Krishna, M. M., Lin, Y. & Englander, S. W. (2002). Cytochrome c folding pathway: Kinetic native-state hydrogen exchange. Proc. Natl. Acad. Sci. U S A, 99, 12173-12178.

56

115. Kim, K. S., Fuchs, J. A. & Woodward, C. K. (1993). Hydrogen exchange identifies native-state motional domains important in protein folding. Biochemistry, 32, 9600-9608.

116. Neira, J. L., Sevilla, P., Menendez, M., Bruix, M. & Rico, M. (1999). Hydrogen exchange in ribonuclease A and ribonuclease S: Evidence for residual structure in the unfolded state under native conditions. J. Mol. Biol. 285, 627-643.

117. Roder, H., Elove, G. A. & Englander, S. W. (1988). Structural characterization of folding intermediates in cytochrome c by H-exchange labelling and proton NMR. Nature, 335, 700-704.

118. Radford, S. E., Dobson, C. M. & Evans, P. A. (1992). The folding of hen lysozyme involves partially structured intermediates and multiple pathways. Nature, 358, 302-307.

119. Mullins, L. S., Pace, C. N. & Raushel, F. M. (1997). Conformational stability of ribonuclease T1 determined by hydrogen-deuterium exchange. Protein Sci. 6, 1387-1395.

120. Perrett, S., Clarke, J., Hounslow, A. M. & Fersht, A. R. (1995). Relationship between equilibrium amide proton exchange behavior and the folding pathway of barnase. Biochemistry, 34, 9288-9298.

121. Clarke, J., Hounslow, A. M., Bycroft, M. & Fersht, A. R. (1993). Local breathing and global unfolding in hydrogen exchange of barnase and its relationship to protein folding pathways. Proc. Natl. Acad. Sci. USA, 90, 9837-9841.

122. Neira, J. L., Itzhaki, L. S., Otzen, D. E., Davis, B. & Fersht, A. R. (1997). Hydrogen exchange in chymotrypsin inhibitor 2 probed by mutagenesis. J. Mol. Biol. 270, 99-110.

123. Chi, Y. H., Kumar, T. K., Kathir, K. M., Lin, D. H., Zhu, G., Chiu, I. M. & Yu, C. (2002). Investigation of the structural stability of the human acidic fibroblast growth factor by hydrogen-deuterium exchange. Biochemistry, 41, 15350-15359.

124. Schmid, F. X. (1990). Spectral methods of characterizing protein conformation and conformational changes. In Protein structure, a practical approach (Creighton, T.E., ed.), pp. 251-285, IRL Press, Oxford.

125. Royer, C. A. (1995). Fluorescence spectroscopy. In Protein stability and folding (Shirley, B. A., ed.), pp. 65-89, Humana Press, Totowa, New Jersey.

126. Eriksson, L., Johansson, E., Kettaneh-Wold, N. & Wold, S. (2001). Multi- and Megavariate Data Analysis Principles and Applications. Umetrics academy.

57

127. Sahlman, L. & Jonsson, B. H. (1992). Purification and properties of the mercuric-ion-binding protein MerP. Eur. J. Biochem. 205, 375-381.

128. Eriksson, P. O. & Sahlman, L. (1993). 1H NMR studies of the mercuric ion binding protein MerP: sequential assignment, secondary structure and global fold of oxidized MerP. J. Biomol. NMR. 3, 613-626.

129. Qian, H., Sahlman, L., Eriksson, P. O., Hambraeus, C., Edlund, U. & Sethson, I. (1998). NMR solution structure of the oxidized form of MerP, a mercuric ion binding protein involved in bacterial mercuric ion resistance. Biochemistry, 37, 9316-9322.

130. Lindberg, M. O., Tangrot, J., Otzen, D. E., Dolgikh, D. A., Finkelstein, A. V. & Oliveberg, M. (2001). Folding of circular permutants with decreased contact order: general trend balanced by protein stability. J. Mol. Biol. 314, 314, 891-900.

131. Otzen, D. E. & Oliveberg, M. (2002). Conformational plasticity in folding of the split beta-alpha-beta protein S6: evidence for burst-phase disruption of the native state. J. Mol. Biol. 317, 613-627.

132. Silow, M., Tan, Y. J., Fersht, A. R. & Oliveberg, M. (1999). Formation of short-lived protein aggregates directly from the coil in two-state folding. Biochemistry, 38, 13006-13012.

133. Villegas, V., Martinez, J. C., Aviles, F. X. & Serrano, L. (1998). Structure of the transition state in the folding process of human procarboxypeptidase A2 activation domain. J. Mol. Biol. 283, 1027-1036.

134. Chiti, F., Taddei, N., White, P. M., Bucciantini, M., Magherini, F., Stefani, M. & Dobson, C. M. (1999). Mutational analysis of acylphosphatase suggests the importance of topology and contact order in protein folding. Nat. Struct. Biol. 6, 1005-1009.

135. van Nuland, N. A., Meijberg, W., Warner, J., Forge, V., Scheek, R. M., Robillard, G. T. & Dobson, C. M. (1998). Slow cooperative folding of a small globular protein HPr. Biochemistry, 37, 622-637.

136. Silow, M. & Oliveberg, M. (1997). Transient aggregates in protein folding are easily mistaken for folding intermediates. Proc. Natl. Acad. Sci. U S A, 94, 6084-6086.

137. Villegas, V., Azuaga, A., Catasus, L., Reverter, D., Mateo, P. L., Aviles, F. X. & Serrano, L. (1995). Evidence for a two-state transition in the folding process of the activation domain of human procarboxypeptidase A2. Biochemistry, 34, 15105-15110.

138. Myers, J. K., Pace, C. N. & Scholtz, J. M. (1995). Denaturant m values and heat capacity changes: relation to changes in accessible surface areas of protein unfolding. Protein Sci. 4, 2138-2148.

58

139. Reimer, U., Scherer, G., Drewello, M., Kruber, S., Schutkowski, M. & Fischer, G. (1998). Side-chain effects on peptidyl-prolyl cis/trans isomerisation. J. Mol. Biol. 279, 449-460.

140. Aronsson, G., Brorsson, A. C., Sahlman, L. & Jonsson, B. H. (1997). Remarkably slow folding of a small protein. FEBS Lett. 411, 359-364.

141. Wand, A. J. & Englander, S. W. (1986). Two-dimensional 1H NMR studies of cytochrome c: assignment of the N-terminal helix. Biochemistry, 25, 1100-1106.

142. Koradi, R., Billeter, M. & Wuthrich, K. (1996). MOLMOL: a program for display and analysis of macromolecular structures. J. Mol. Graph. 14, 32-51.

143. Avbelj, F., Luo, P. & Baldwin, R. L. (2000). Energetics of the interaction between water and the helical peptide group and its role in determining helix propensities. Proc. Natl. Acad. Sci. U S A, 97, 10786-10791.

144. Avbelj, F. & Baldwin, R. L. (2002). Role of backbone solvation in determining thermodynamic beta propensities of the amino acids. Proc. Natl. Acad. Sci. U S A, 99, 1309-1313.

145. Makhatadze, G. I. & Privalov, P. L. (1995). Energetics of protein structure. Adv. Protein Chem. 47, 307-425.

146. Steele, R. A. & Opella, S. J. (1997). Structures of the reduced and mercury-bound forms of MerP, the periplasmic protein from the bacterial mercury detoxification system. Biochemistry, 36, 6885-6895.

147. Kjellsson, A. (2002). Characterisation of partly folded states in proteins using hydrogen exchange. Thesis, Department of biochemistry, Umeå.

148. Silow, M. & Oliveberg, M. (1997). High-energy channeling in protein folding. Biochemistry, 36, 7633-7637.

149. Ternstrom, T., Mayor, U., Akke, M. & Oliveberg, M. (1999). From snapshot to movie: phi analysis of protein folding transition states taken one step further. Proc. Natl. Acad. Sci. USA, 96, 14854-14859.