UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Mechanical denaturation : forced unfolding of proteins Li, Yongnan 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2013_spring_li_yongnan.pdf [ 4.59MB ]
Metadata
JSON: 24-1.0073778.json
JSON-LD: 24-1.0073778-ld.json
RDF/XML (Pretty): 24-1.0073778-rdf.xml
RDF/JSON: 24-1.0073778-rdf.json
Turtle: 24-1.0073778-turtle.txt
N-Triples: 24-1.0073778-rdf-ntriples.txt
Original Record: 24-1.0073778-source.json
Full Text
24-1.0073778-fulltext.txt
Citation
24-1.0073778.ris

Full Text

Mechanical Denaturation: Forced Unfolding of Proteins by Yongnan Li B.Sc., The University of British Columbia, 2010 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in The Faculty of Graduate Studies (Chemistry) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2013 c© Yongnan Li 2013 Abstract Mechanical denaturation has emerged as a novel method to study chemical and physical properties of protein molecules. In this thesis, single-molecule force spectroscopy has been carried out using the atomic force microscope to investigate the mechanical design of proteins through denaturation via an applied mechani- cal force. In the first study, a small globular protein has been shown to exhibit pronounced anisotropic response to directional mechanical stress. One protein can be both mechanically strong and weak. It will be strong when direction of the force vector is aligned with particular structural elements of the protein, and it will be weak otherwise. Mechanical denaturation in the strong direction is ac- companied by cooperative disruption of intramolecular interactions in the protein. Conversely, mechanical denaturation in the weak direction is accompanied by se- quential disruption of those same interactions. In the second study, the mechanical properties of a cofactor dependent protein is characterized. It is shown that both the protein and cofactor are mechanically strong in the presence of the cofactor. Removal of the cofactor tremendously diminishes the mechanical strength of the protein. The mutually supportive roles of structure and function are demonstrated through mechanical denaturation experiments. ii Preface The work presented in this thesis is based on results of collaborative research. Contributions by other collaborators and/or co-authors are stated below. Chapter 3 has been published as: “Li, Y. D., Lamour, G., Gsponer, J., Zheng, P. & Li, H. B. The Molecular Mechanism Underlying Mechanical Anisotropy of the protein GB1. Biophys. J. 103, 2361-2368 (2012).” My supervisor, Hongbin Li, designed the research. Peng Zheng helped me with protein engineering work. Guillaume Lamour helped me with SMD simulation work. I performed the rest of the experiments and simulations. I wrote the manuscript. All of the authors discussed the results and edited the manuscript. This chapter was based on the published manuscript, but was written by the author unless otherwise noted. Chapter 4 is based on unpublished work. My supervisor, Hongbin Li, de- signed the research. Peng Zheng helped me with molecular biology and protein engineering work. I performed the rest of the experiments and wrote the chapter. iii Table of Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii List of Symbols and Abbreviations . . . . . . . . . . . . . . . . . . . . . x Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Energy landscape and protein folding . . . . . . . . . . . . . . . 2 1.1.1 Protein structure and folding . . . . . . . . . . . . . . . . 2 1.1.2 Energy landscape . . . . . . . . . . . . . . . . . . . . . 3 1.2 Mechanical denaturation of a protein . . . . . . . . . . . . . . . 6 1.2.1 Experimental techniques for mechanical denaturation . . 6 1.2.2 Entropic polymer elasticity . . . . . . . . . . . . . . . . 10 1.2.3 Two-state kinetic model for mechanical denaturation . . . 12 1.3 Molecular determinants of mechanical strength . . . . . . . . . . 16 1.3.1 Folding topology . . . . . . . . . . . . . . . . . . . . . . 16 1.3.2 Cofactor binding . . . . . . . . . . . . . . . . . . . . . . 18 iv Table of Contents 2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.1 Protein engineering . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.2 Coupling reaction . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.3 Instrumentation of AFM . . . . . . . . . . . . . . . . . . . . . . 21 2.3.1 Tip-cantilever assembly . . . . . . . . . . . . . . . . . . 22 2.3.2 Piezoelectric actuator . . . . . . . . . . . . . . . . . . . 24 2.3.3 Control system . . . . . . . . . . . . . . . . . . . . . . . 24 2.3.4 Isolation and noise reduction . . . . . . . . . . . . . . . 25 2.4 Force spectroscopy with AFM . . . . . . . . . . . . . . . . . . . 25 2.4.1 Cantilever spring constant calibration . . . . . . . . . . . 25 2.4.2 Contour length increment . . . . . . . . . . . . . . . . . 26 2.4.3 Rupture force statistics . . . . . . . . . . . . . . . . . . . 29 2.4.4 Monte Carlo simulation procedure . . . . . . . . . . . . 29 2.5 SMD simulations . . . . . . . . . . . . . . . . . . . . . . . . . . 31 3 Molecular Mechanism Underlying Mechanical Anisotropy . . . . . 32 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.2 Experimental section . . . . . . . . . . . . . . . . . . . . . . . . 34 3.2.1 Protein engineering . . . . . . . . . . . . . . . . . . . . 34 3.2.2 Thiol-maleimide coupling reaction . . . . . . . . . . . . 35 3.2.3 Single-molecule AFM experiments . . . . . . . . . . . . 35 3.2.4 SMD simulations . . . . . . . . . . . . . . . . . . . . . 36 3.3 Design of loading directions . . . . . . . . . . . . . . . . . . . . 37 3.4 Anisotropic mechanical response of GB1 . . . . . . . . . . . . . 39 3.5 Mechanical unfolding mechanisms . . . . . . . . . . . . . . . . 44 3.6 Detailed mechanisms from native contact analysis . . . . . . . . 47 3.7 Diverse unfolding mechanisms lead to anisotropy . . . . . . . . . 50 4 Mechanical Unfolding of an Iron-Sulphur Protein . . . . . . . . . . 52 4.1 Experimental section . . . . . . . . . . . . . . . . . . . . . . . . 53 4.1.1 Protein engineering . . . . . . . . . . . . . . . . . . . . 53 v Table of Contents 4.1.2 Thiol-maleimide coupling reaction . . . . . . . . . . . . 54 4.1.3 Single-molecule AFM experiments . . . . . . . . . . . . 54 4.2 Structure of a plant-type [2Fe2S] ferredoxin . . . . . . . . . . . . 54 4.3 Mechanical design of spinach ferredoxin . . . . . . . . . . . . . 58 4.4 Rupture of protein structure and [2Fe2S] cluster under force . . . 60 5 Conclusions and Future Prospects . . . . . . . . . . . . . . . . . . . 66 Appendix A Additional UV-Visible Spectra . . . . . . . . . . . . . . . . . . . . . 67 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 vi List of Tables 2.1 Restriction endonuclease digestion sites used in this thesis . . . . 20 3.1 Kinetic data on mechanical unfolding of GB1 constructs . . . . . 43 vii List of Figures 1.1 Energy landscape of a single protein molecule in vitro . . . . . . . 4 1.2 The protein folding reaction . . . . . . . . . . . . . . . . . . . . 5 1.3 Constant velocity mode of AFM . . . . . . . . . . . . . . . . . . 7 1.4 Mechanical denaturation of a hetero-dimeric polyprotein . . . . . 9 1.5 Illustration of an ideal chain . . . . . . . . . . . . . . . . . . . . 11 1.6 Effect of force on the free energy surface . . . . . . . . . . . . . . 13 1.7 Denaturation probability as a function of force . . . . . . . . . . . 15 1.8 Folding topology and mechanical strength . . . . . . . . . . . . . 17 1.9 Cofactor binding and mechanical strength . . . . . . . . . . . . . 19 2.1 Schematic diagram of an AFM . . . . . . . . . . . . . . . . . . . 21 2.2 Cantilever SEM image . . . . . . . . . . . . . . . . . . . . . . . 23 2.3 Contour length increment as single-molecule fingerprint . . . . . 28 2.4 Information from rupture force statistics . . . . . . . . . . . . . . 30 3.1 Ribbon cartoon representation of the structure of GB1 . . . . . . . 38 3.2 Anisotropic response of GB1 to mechanical stress . . . . . . . . . 40 3.3 Dependency of unfolding force on the pulling velocity . . . . . . 42 3.4 Force versus extension plots with snapshots . . . . . . . . . . . . 45 3.5 Fraction of native contacts between structural elements . . . . . . 48 4.1 Representation of the structure of FdS with iron-sulphur cluster . . 55 4.2 Biogenesis of iron-sulphur proteins in vivo . . . . . . . . . . . . . 56 4.3 UV-visible absorption spectrum of spinach ferredoxin . . . . . . . 57 4.4 Coordination sphere of iron in the cluster . . . . . . . . . . . . . 59 viii List of Figures 4.5 Mechanical design of FdS . . . . . . . . . . . . . . . . . . . . . 61 4.6 UV-visible absorption spectrum of cF93c . . . . . . . . . . . . . 63 4.7 Iron-sulphur cluster and mechanical design of FdS . . . . . . . . 64 A.1 UV-visible absorption spectrum of concentrated cF93c . . . . . . 67 A.2 UV-visible absorption spectra . . . . . . . . . . . . . . . . . . . . 68 A.3 UV-visible absorption spectra in the presence/absence of EDTA . 69 ix List of Symbols and Abbreviations a constant, rate of change for force as a linear function of time Å Ångström (10−10 metre) aa amino acid(s) or amino acid residue(s) AFM Atomic Force Microscopy or Atomic Force Microscope C2A first C2 domain of synaptotagmin I EDTA ethylenediaminetetraacetic acid F(x), F force as a function of polymer extension, force FdS [2Fe2S] ferredoxin of type I from spinach GB1 B1 immunoglobulin G binding domain of streptococcal protein G I27 27th immunoglobulin-like domain of the protein titin I28 28th immunoglobulin-like domain of the protein titin InvOLS inverse optical lever sensitivity kB Boltzmann’s constant ks spring constant of an AFM cantilever ku(F), ku unfolding rate constant as a function of force k◦u intrinsic unfolding rate constant Lc contour length of polymer in WLC model ∆Lc contour length increment MC Monte Carlo nm nanometre (10−9 metre) p persistence length of polymer in WLC model PN fraction of population in native state or survival probability PD fraction of population in native state or denaturation probability x List of Symbols and Abbreviations PDB protein data bank pN pico-Newton (10−12 Newton) SD standard deviation SEM scanning electron microscope SMD steered molecular dynamics SMFS single-molecule force spectroscopy t time T absolute temperature in Kelvin WLC Worm-Like Chain model of polymer elasticity wt wild type x extension of polymer xc AFM cantilever deflection in nm xp AFM piezoelectric actuator movement in nm xu unfolding distance [2Fe2S] iron sulphur cluster with 2 iron and 2 sulphur atoms ∼ on the order of or similar to ∆GTS-D free energy barrier for folding ∆GTS-N free energy barrier for unfolding xi Acknowledgements I would like to thank my supervisor, Dr. Hongbin Li, and all past and present mem- bers of the research group for their help and friendship. I would also like to thank our collaborator Dr. Jörg Gsponer and all members of his research group. I thank Dr. Elliott Burnell for critical reading of the thesis and his comments/suggestions on the thesis. Finally, I would like to acknowledge financial support from the Nat- ural Sciences and Engineering Research Council of Canada and the University of British Columbia. xii Chapter 1 Introduction Advent of new experimental methods often follow great leaps in technology. The development of the atomic force microscope (AFM) by Gerd Binnig and col- leagues in the 1980s [1], shortly after the Nobel prize winning design of the scan- ning tunneling microscope, has brought a versatile new tool into the toolboxes of biochemists and biophysicist [2]. Amongst the various applications of AFM in biophysics, single-molecule force spectroscopy (SMFS) has evolved to become an indispensable technique in the studies of: cellular mechanics and molecular elas- ticity of biopolymers [3–5], mechanical (un)folding of protein [6, 7], and design principles for mechanical strength in protein [8, 9]. Although SMFS is still a relatively young experimental method, the amount of work and volume of literature accumulated over the years still make the task of providing a comprehensive review of the whole field nearly impossible. Thus, I have elected to present an overview of SMFS as applied to the studies of protein (un)folding and mechanical design of proteins. The rest of the introduction is organized into three sections: 1. Energy landscape and protein folding, 2. Mechanical denaturation of a protein, 3. Molecular determinants of mechanical strength. 1 1.1. Energy landscape and protein folding 1.1 Energy landscape and protein folding 1.1.1 Protein structure and folding Flip open any modern introductory molecular biology [10] or biochemistry [11] textbooks, one is bound to find large sections dedicated to topics on: protein, protein structure, and protein function. Proteins, encoded by nucleic acids which carry the genetic information, are often referred to as workhorses in the cell. Pro- teins can serve as the key structural components (e.g. tubulins in microtubules); they can act as enzymes to facilitate chemical transformations (e.g. restriction en- donuclease); they can behave like sensors to carry out bio-recognition processes (e.g. human antibodies). How could one class of molecules carry out such di- verse array of functions? The answer to this question lies in the structure of the molecule. Proteins are linear polymers formed from amino acids (aa) via covalent pep- tide linkages between adjacent neighbours. Each protein has its specific sequence of aa and a single chain of aa is often referred to as a polypeptide. The reader is referred to textbooks on molecular biology [10] and/or biochemistry [11] for the basic chemical and physical properties of amino acids and polypeptides. In order to perform their destined functions in vivo, most proteins need to adopt one or a few particular conformation (i.e. native state of the protein) in three-dimensional space. The native state of the protein is typically a highly ordered state, with well-defined atomic contacts between various parts of the polypeptide chain. The term protein structure is designated to describe these contacts in the native state and protein structure can be described in a hierarchy. The sequence of aa in a polypeptide chain determine the primary structure of the protein. The secondary structure (e.g. α-helix or β-sheet) results from backbone hydrogen bonds between hydrogen attached to one amide nitrogen and carbonyl oxygen of another amide in the polypeptide chain, whereas tertiary structure captures atomic level contacts between parts of protein proximate in space but distant in aa sequence. Structural information about a protein can be extracted at atomic resolution with powerful 2 1.1. Energy landscape and protein folding experimental methods like X-ray crystallography and nuclear magnetic resonance spectroscopy. Knowledge of protein structure most often serve as a great starting point for the study of protein folding. The problem of protein folding focuses on providing an explanation as to how does a linear polypeptide chain transit from a highly disordered state to the native state and become a functional protein. Moreover, the folding process needs to be explained on a relevant time scale. Monumental work carried out by Anfinsen and many others have demonstrated that the primary structure of the polypeptide contains all the information required to fold and reach the native state for most proteins [12]. This is such an amazing result, considering the astronomical num- ber of accessible conformations. Although the peptide bonds in the chain impose certain restrictions on its internal rotations, the polypeptide backbone still has sig- nificant conformational freedom. Moreover, side-chains of aa in the polypeptide have even less steric hindrance and add a large number of degrees of freedom to the conformation space. The problem of protein folding can be now rephrased in the following way: how does the polypeptide chain explore its conformation space such that it finds its native state on a biologically relevant time scale. 1.1.2 Energy landscape The currently accepted view of protein folding is the funnel-like energy landscape view [14, 15]. Levinthal’s paradox is often invoked to argue against the possibility of protein folding being a mere random search over conformation space. Inter- ested readers are referred to a reference for a detailed discussion of Levinthal’s paradox and its resolution under a guided search scenario [16]. From the atomistic point of view, protein resides in a noisy environment with constant non-stopping bombardment from solute and solvent molecules. Diffusive motion driven by thermal fluctuation dominates on this size scale and serves as the driving force for protein folding. Under thermal fluctuation, the protein samples many different conformation with many different atomic contacts either native or non-native. On average, native atomic contacts as in the native state of the 3 1.1. Energy landscape and protein folding Figure 1.1: Energy landscape of a single protein molecule in vitro. Schematic diagrams illustrate how funnel-like energy landscapes guide the folding of a sin- gle polypeptide chain. The vertical scale is energy and horizontal scale is con- figuration entropy. Two examples are provided for comparison: smooth funnel (a) versus rugged funnel (b). Energy scale is on the order of 1-100 kBT (see Fig. 1.2 panel b). Adapted by permission from Macmillan Publishers Ltd: Nature Structural and Molecular Biology [13] c©(2009). protein are more favourable since these tend to lower the energy. The energy scale is on the order of 1-100 kBT (see Fig. 1.2 panel b). These native atomic contacts persist longer than non-native ones and guide the protein to the lowest energy conformation (i.e. the native state). One way of describing the situation in the language of mathematics is by considering energy as a scalar function of conformation. In this senario, the energy hypersurface will be embedded in the conformation space and the native state typically will be the global minima on this high-dimensional surface. This is commonly referred to as the energy landscape for protein folding. The energy landscape is best understood with graphical illustrations (see Fig. 1.1). Out of all the accessible conformations, unfolded polypeptides at the very top of the funnel can adopt an astronomical number of conformations which are high in energy. Various types of intramolecular and intermolecular interactions limits the conformation freedom of the polypeptides, but these contacts (driven by 4 1.1. Energy landscape and protein folding Figure 1.2: The protein folding reaction. Schematic diagrams illustrate two descriptions of the protein folding reaction: free energy surface (a) versus funnel- like folding energy landscape (b). Panel a: free energy axis has scale of 10 kBT . Panel b: energy axis has scale of 100 kBT and local minimum has scale of 2-3 kBT . Free energy surface in panel a can be viewed as a one dimensional projec- tion of the high-dimensional energy surface onto a particular reaction coordinate (i.e. degree of nativeness). Schematic energy landscape in panel b will be a two dimensional illustration of the high-dimensional energy surface. Route 1 and 2 shown as thin white lines may have different free energy surfaces. Reprinted from [14], c©(2004), with permission from Elsevier. hydrophobic interactions, hydrogen bonds, electrostatic interactions etc.) In our scenario, configuration entropy represents number of accessible conformations of the polypeptide chain. The native state is equivalent to one or possible a few well-defined conformations with relatively low energy. The native state serves as a trap/sink in the funnel, towards which the polypeptide chain will be driven by thermal fluctuation. Although the native state is the only trap in a smooth funnel (Fig. 1.1 a), this is not the only case. Intermediates and other kinetic traps of various depth makes a rugged funnel (Fig. 1.1 b) much more difficult to navigate. 5 1.2. Mechanical denaturation of a protein 1.2 Mechanical denaturation of a protein Since most proteins assume their native states in vivo to perform their function, it is sensible that the native state is typically quite stable under native condition (i.e. appropriate pH, ionic strength, viscosity, etc.). To study the protein folding problem in vitro, experimentalists need to perturb the protein in question from its native state using denaturation techniques. Classic ensemble denaturation ex- periments involve the use of heat or chemical denaturant and have laid down the cornerstone of our understanding in protein folding. Results from many classi- cal studies using these techniques have entered textbooks of molecular biology, biochemistry, and physical chemistry. In comparison, mechanical denaturation is still in its adolescence. The idea that mechanical force could act as a denaturing agent only has emerged and became possible, as advances in technology made nano-manipulation techniques accessible. The question is: why would anyone be interested in using force to denature a protein? To answer this question, we need to consider the role of mechanical force in biology. For advanced cells like eukaryotic cells, mechanical processes are present in almost all essential cellular processes including: replication, transcription, translation of the genetic material, translocation and degradation of proteins, and cell locomotion and signalling [17– 19]. 1.2.1 Experimental techniques for mechanical denaturation Over the past three decades, a number of experimental techniques for mechanical denaturation have been developed [20]. Optical tweezers, magnetic tweezers, and force mode AFM are three examples of such techniques. In order to lay down a good comparison of these techniques, modes of operation and physical principle of each technique need to be explained in detail which is not the aim of this thesis. Readers of this thesis are kindly referred to a review article [21] which provide a very concise and readable overview and comparison of these three techniques. The rest of this section will focus on explaining principles of operation behind 6 1.2. Mechanical denaturation of a protein Figure 1.3: Constant velocity mode of AFM. Schematic diagrams illustrate a typical mechanical denaturation experiment carried out with the AFM. Mechan- ical denaturation of the protein proceeds in four steps: (1) Protein molecule is attached to both the AFM tip and substrate surface. (2) Substrate moves away from the AFM tip at a constant velocity, applying a force to the protein. (3) Force promotes the denaturation of the protein, one of the monomeric domain in the lin- ear oligomer unfolds and force is reduced. (4) Substrate continues to move away, steps (2) and (3) repeat. Red circles highlight the rupture force at each unfolding event and the blue circle highlights the detachment force of the protein from either AFM tip side or substrate surface side. force mode AFM, since it is the instrument used to conduct research presented in latter chapters of this thesis. The AFM is initially developed as a tool to image materials on the atomic scale beyond the diffraction limit of light [1]. Soon after its invention, the capabilities of the AFM to measure molecular scale interaction forces (∼pN) on nanometer spatial scale (lateral ∼nm and vertical ∼Å) are realized and exploited [22, 23]. The operational principles behind the AFM will be explained in the Materials and Methods section 2.3 of the thesis. In this section, I will focus on explaining how 7 1.2. Mechanical denaturation of a protein can AFM be used to perform mechanical denaturation of protein. The AFM uses a micro-fabricated cantilever equipped with a sharp tip as the mechanical probe. When a protein molecule is attached to both the AFM tip and the glass substrate through non-specific interactions, the molecule can be stretched via controlled motion of the positioner. Proteins used in these exper- iments are most often linear oligomer made of identical monomeric domains co- valently linked to each other (often called a polyprotein, rationales for using such constructs are provided in Materials and Methods subsection 2.4.2). When the separation distance between the AFM tip and substrate mounted on the positioner, the protein molecule is subjected to mechanical stress. Under the most common experimental protocol, the separation distance is increased as a linear function of time and the protein is extended accordingly. Force versus extension profile of a single polyprotein shows a characteristic sawtooth pattern with equally spaced peaks each corresponding to the denaturation of one domain of the polyprotein. (see Fig. 1.3 and Fig. 1.4). How can we know the above interpretation of experimental data is the correct way to do it? Since the protein sample is deposited on a substrate surface, there is always the possibility of adsorption and the equally space force peaks correspond to sequential desorption off the surface rather than mechanical denaturation. The giant muscle protein titin has been a favourite subject of study in mechanical de- naturation [25, 26]. Titin contains many individually folded immunoglobulin-like domains. Each of these domains contains about 90 aa and can be individually produced and isolated using molecular biology techniques. Questions concerning the interpretation of experimental data have been addressed by Li and colleagues through a clever design in the polyprotein [27]. The 27th (I27 for short) and 28th (I28) immunoglobulin-like domains of titin were first characterized individually and found to show rupture forces of ∼200 pN and ∼300 pN respectively. Through the construction of a hetero-dimeric polyprotein with alternating I27 and I28 do- mains, Li and colleagues have demonstrated rupture events are sorted according to force rather than sequence of appearance in the polyprotein. This finding essen- 8 1.2. Mechanical denaturation of a protein Figure 1.4: Mechanical denaturation of a hetero-dimeric polyprotein. A polyprotein molecule [I27-I28]4 is constructed based on a hetero-dimeric I27- I28 construct (A). Mechanical denaturation experiment shows the mechanically weaker I27 domain always denatures before the mechanically stronger I28 do- main, regardless of their sequence of appearance in the polyprotein (B). This offers strong evidence against the argument that force peaks may be due to sequential surface desorption. Reprinted from [24], c©(2000), with permission from Elsevier. tially has ruled out sequential desorption off the surface as a plausible explanation of the experimental data. Under typical experimental conditions for constant velocity mechanical de- naturation experiments, the unfolding of a protein is a non-equilibrium process. If extension of the polyprotein stops before the detachment and the molecule is relaxed to small extension, one will observe the retraction trace will not overlap with the extension trace [28]. The area enclosed by the extension and retraction traces represent irreversible work done in the mechanical unfolding process. Note the fact that we do not have direct control over force in the constant ve- locity scheme, the force versus extension behaviour of the protein is governed by its elastic properties. It is important to emphasize almost all mechanical denatura- tion experiments are carried out under thermal-activated regime [29] in which the applied force act to activate the unfolding reaction by lowering the energy barrier 9 1.2. Mechanical denaturation of a protein for mechanical unfolding. A more detailed explanation of the role of force on the unfolding reaction is provided in section 1.2.3. 1.2.2 Entropic polymer elasticity In order to gain a better understanding of the mechanical denaturation process, some elementary ideas from polymer physical chemistry are required. When com- pared to SMFS, the field of polymer physical chemistry is much more mature with prodigous amount of work carried out by eminent scientists like P. J. Flory. In the language of polymer physical chemistry, a polymer is a linear chain molecule made from monomeric units. An ideal chain is the simplest model for a polymer, since relative orientations of neighbouring monomers are completely independent and all interactions are neglected (see Fig. 1.5 for an illustration). The length of such a polymer can be measured in two ways. First, the end-to-end length is simply the Euclidean distance measured from the spatial coordinates of the two ends of the polymer. Second, the contour length of the polymer is calculated by summing the length of individual monomers over the number of monomers in one polymer. From the definitions, one can see the contour length is a property of the polymer defined by the number and nature of the monomers but the end-to-end length depends on the relative orientation of the monomers (i.e. conformation of the polymer). The end-to-end length of an ideal chain in solution will always be smaller than its contour length at finite non-zero temperature. To see why this is the case, consider the following microscopic picture of a polymer in solution: thermal motion is constantly driving the solvent molecules into the polymer and changing the relative orientation of the monomers. Starting with a fully extended ideal chain (end-to-end length equals contour length) without tethering the two ends, constant bombardments from the solvent molecules quickly change the conformation of the polymer and drive it to sample its conformation space. All conformations of the polymer are equally likely to be occupied because they all have the same energy. Since the fully extended linear conformation is only one conformation out of the 10 1.2. Mechanical denaturation of a protein Figure 1.5: Illustration of an ideal chain. Vector representation illustrates the ideal chain model of a polymer. Here ~ri is monomer i and contour length of the polymer is given by the sum over length of all monomers. ~R is the end to end vector, whose length is the end to end length. enormous number of possible conformations, the polymer spends most of its time sampling one of the many more compact conformations. It is well known that even for an ideal chain, a non-zero force is needed to extend it. To better understand this, we briefly go back to the microscopic view of last paragraph. In order to extend a polymer in solution to its contour length, work must be done to counteract the constant bombardment of solvent molecules. This energy can be supplied in the form of an applied force. The force versus extension behaviour of various polymer models have been analyzed and presented with great detail in many books on polymer physics and physical chemistry [30, 31]. It has been shown in experiments that the force versus extension behaviour of the protein molecule is described reasonably well by the worm-like chain (WLC) model of polymer elasticity [25]. WLC model describes the polymer as a thin continuous elastic filament whose bending energy obeys Hooke’s law of elastic- ity [31]. Unlike the ideal chain model in which relative orientations between monomers are completely un-correlated, orientation correlation in WLC model is lost on length scale larger than the persistence length. In the case of WLC model, no closed-form expression of force as function of extension is known but a nu- 11 1.2. Mechanical denaturation of a protein merical solution exists. An interpolation formula is proposed to aid the analysis of experimental data: F(x) = kBT p [ 1 4 1 (1 − x/Lc)2 − 1 4 + x Lc ] for 0 ≤ x < Lc, (1.1) where F(x) is force as a function of polymer extension, kB is Boltzmann’s constant, T is absolute temperature, p is persistence length of polymer in WLC model, x is extension of polymer, Lc is contour length of polymer in WLC model. 1.2.3 Two-state kinetic model for mechanical denaturation Many proteins fold in a highly cooperative fashion, which is well described by the two-state kinetic model [33]. Thus, two-state kinetic model for protein fold- ing serves as an ideal venue for introducing the effect of force into protein fold- ing/unfolding model (see Fig. 1.6 for an illustration). Under the two-state model, the protein molecule can exists in one of two states: native state and denatured state. Transition between the two states go through a single transition state. The unfolding rate constant for transiting from the native state to the denatured state is given by: k◦u = A exp (−∆GTS−N kBT ) , (1.2) where k◦u is intrinsic unfolding rate constant, A is an Arrhenius type pre-exponential factor, ∆GTS−N is free energy barrier for unfolding. 12 1.2. Mechanical denaturation of a protein Figure 1.6: Effect of force on the free energy surface. Schematic diagram illus- trates the effect of force on the free energy surface for protein folding/unfolding reaction. Vertical axis is the Gibbs free energy. Horizontal axis is the reaction coordinate. Gray curve is the original surface and black curve is shifted by a force which depends linearly on the reaction coordinate. Labels N, D, and TS denote the native state, denatured state, and transition state, respectively. Labels ∆GTS−N and ∆GTS−D denote the original free energy barriers for unfolding and folding, respectively. Label xu denotes the unfolding distance between transition state and native state, whereas x f denotes the folding distance between transition state and denatured state. Reprinted from [32], c©(2004), with permission from Elsevier. In the field of SMFS, phenomenological model popularized by Bell is one of the simplest and the most often invoked model for incorporating effect of an applied force in kinetics [34]. 13 1.2. Mechanical denaturation of a protein ku (F) = A exp (−∆GTS−N + xuF kBT ) = k◦u exp ( xuF kBT ) , (1.3) where ku(F) is unfolding rate constant as a function of force, F is applied force, xu is unfolding distance between transition state and native state. The exponential term on the right hand side of equation 1.3 implies the free en- ergy barrier for unfolding is lowered by Fxu (see Fig. 1.6). Under influence of force, transition state TS and denature state D is lowered by Fxu and F(xu + x f ), respectively. This means the free energy barrier for unfolding is lowered by Fxu, whereas the free energy barrier for folding is raised by Fx f . A similar expres- sion can be derived for k f (F), the folding rate constant. For a more theoretically ground discussion of the justifications and limitations of such a simple model, the readers of this thesis is kindly referred to the references [35–38]. Kinetics of denaturation under the two-state model can be described as a first order reaction. The simplest case is often the most illuminating. If the folding rate constant is assumed to be zero (k f = 0) and force is assumed to be a linear function of time (F = at, a is constant), the rate equation can be solved easily. Under these assumptions, the rate equation is: − dPN(t) dt = ku(F(t))PN(t) = k◦u exp ( xuF(t) kBT ) PN(t), (1.4) where PN(t) is fraction of population in native state (N) as a function of time, F(t) is force as a function of time. Assuming all of the population is in the native state (PN = 1) at the beginning (t = 0), solution to equation 1.4 is: PN(t) = exp [ k◦u kBT xua ( 1 − exp ( xuat kBT ))] . (1.5) 14 1.2. Mechanical denaturation of a protein Fraction of population in denatured state PD is given by PD = 1 − PN . It can be written as a function of F, with the substitution F = at, as follows: PD(F(t)) = 1 − exp [ k◦u kBT xua ( 1 − exp ( xuF(t) kBT ))] . (1.6) Figure 1.7: Denaturation probability as a function of force. Plots are generated using equations 1.6 and 1.7 with WolframAlpha R© (wolframalpha.com). Mechanical denaturation of single protein molecule is a stochastic process. On the single-molecule level, PD is the denaturation probability and PD(F) can be view as the cumulative distribution function. The derivative of denaturation probability with respective to force (dPD/dF) will yield the probability density function: dPD dF = k◦u a exp [ k◦u kBT xua ( 1 − exp ( xuF(t) kBT ) + xuF(t) kBT )] . (1.7) The significance of the probability density function will be discussed when it will be used in subsection 2.4.3 (see Fig. 1.7 for typical behaviours of the cu- mulative distribution function and probability density function). However, it is important to note that the assumptions made in the begin for the derivations are not necessarily fulfilled in a typical experiment. In a constant velocity AFM ex- periment, it is extension not force that is a linear function of time. The dependence 15 1.3. Molecular determinants of mechanical strength of force on extension is given by equation 1.1. In such a case, we can anticipate that the derivation of the probability density function will not be trivial. 1.3 Molecular determinants of mechanical strength Following pioneering works in the late 1990s [25, 39, 40], the field of SMFS has grown tremendously. Ever since the first experiment performed on the giant mus- cle protein titin, the determination of molecular origins of mechanical strength has become one of the focuses of the field [8, 24]. In the simplest sense, mechanical strength or stability of a protein can be defined as the ability to resist an applied force before unfolding. Here I would like to present a brief overview of current state of knowledge on structural design principles behind molecular mechanical strength of a single protein. To review results in the field that are most relevant to the work presented in this thesis, the rest of this section is divided into two parts: 1. Folding topology, 2. cofactor binding. 1.3.1 Folding topology Topology of a folded protein refers to the relative orientations of various sec- ondary structural elements in three dimensional space. Very early on, researchers have realized that α-helical proteins are mechanically weak while β-sheet pro- teins are mechanically strong. The immunoglobulin-like domain (e.g. I27) from the giant muscle titin, adopting a so-called β-sandwich fold, displays significant mechanical strength [25]. Cytoskeletal protein spectrin, adopting a so-called α- helix bundle fold, displays low mechanical strength [41]. However, it is not the type of secondary structure per se that determine the mechanical strength. Combinations of results from SMFS and steered molecular 16 1.3. Molecular determinants of mechanical strength Figure 1.8: Folding topology and mechanical strength. Panel (a) top, force versus extension trace of a linear oligomer (I27)n (typical n = 8 to 12) of individ- ual I27 domains. Average height of force peaks is ∼200 pN. Thin blue curves are generated by fitting successive WLC force versus extension curves using equa- tion 1.1. Panel (a) bottom, ribbon cartoon representation of protein I27 structure with backbone hydrogen bonds represented as black bars. Schematic diagram of β-strands G and A’ with hydrogen bonds as dashed lines, applied force acts on the β-strands in a shearing geometry. Panel (b) top, force versus extension trace of (C2A)n. Average height of force peaks is ∼60 pN. Panel (b) bottom, ribbon cartoon representation of protein C2A structure with backbone hydrogen bonds represented as black bars. Schematic diagram of β-strands G and A with hydro- gen bonds as dashed lines, applied force acts on the β-strands in a un-zipping geometry. Reprinted from [24], c©(2000), with permission from Elsevier. dynamics (SMD) show that differences in mechanical strength arise from the dif- ferences in orientation of the secondary structures with respect to the applied force vector [24, 42–44]. When the applied force acts to shear two β-strands which are distant in sequence but directly interacting with each other, higher force is re- quired to unfold a domain. This is thought to be due to the simultaneous rupture of various non-covalent interactions (e.g. hydrogen bonds) existing between the neighbouring β-strands (referred to as mechanical clamp: see Fig. 1.8 panel a). On the other hand, a lower force is required when the applied force act to un-zip 17 1.3. Molecular determinants of mechanical strength two neighbouring β-strands. This is because sequential rupture of non-convalent interactions is thought to happen (cf. Fig. 1.8 panel b). Relative orientation of force vector, with respect to arrangement of structural elements, also depend on direction of the pulling axis. Examples shown in Fig. 1.8 involve two different proteins and both are pulled from the two ends of the polypeptide chain. Recent development in polyprotein construction protocols have enabled a single protein to be pulled through different anchoring residues effectively changing the pulling axis [45, 46]. Results from such studies reveal fascinating anisotropic behaviours from a single protein under mechanical stress. 1.3.2 Cofactor binding Discovery of the relationship between native topology and mechanical stability is encouraging, since mechanical strength of a protein can be related to molecular level structural information. The rational design of mechanical strength based on this knowledge is a good test of our understanding. Since protein to protein and protein to cofactor interactions are ubiquitous in biology, rational tuning of mechanical strength base on these interactions is of great interest. One method in this family (of methods) involves introduction of a metal ion cofactor binding site through protein engineering [47] (see Fig. 1.9). In this study, a bi-histidine motif capable of chelating divalent metal ions (e.g. Ni2+) is intro- duced into the mechanical clamp region of the protein GB1. Mechanical strength of GB1 can be enhanced when cofactor binding selectively stabilize the native state over the mechanical transition state, which effectively increases the free en- ergy barrier to mechanical unfolding. Enhancement to the mechanical strength can be fully reversed with the ad- dition of a competing chelator (e.g. imidazole in [47]) or a much more potent chelator (e.g. EDTA [48]). Cao and colleagues have noted that mixed species can be observed in the same polyprotein molecule when the ion concentration used is much smaller than the saturation concentration [49]. This suggests the character- istic time for cofactor binding is on a similar or slow time scale than typical time 18 1.3. Molecular determinants of mechanical strength Figure 1.9: Cofactor and mechanical strength. Engineered metal ion chelation has been demonstrated to be a powerful method for increasing the mechanical strength of a protein. Bi-histidine motif are introduced across the two terminal β- strands at three different residue pairs (A, D, G). Force versus extension traces (B, E, H) and rupture force histograms (C, F, I) clearly show enhanced mechanical strength after cofactor (i.e. Ni2+) binding. Reprinted from [47] for educational purposes, c©(2008), National Academy of Sciences. scale of the experiment. The metal ion chelated GB1 protein becomes an inter- esting model system for studying the intertwined processes of protein folding and cofactor binding [50]. 19 Chapter 2 Materials and Methods 2.1 Protein engineering Gene encoding proteins of interest are either received as a gift or purchased from a commercial supplier. Single aa mutations are performed with a standard site- directed mtagenesis method. Gene constructs are subcloned (see Table 2.1) into DNA plasmids used for either propagation or expression. All mutant constructs are overexpressed in Escherichia coli and purified using affinity chromatography. BamHI BglII KpnI digestion 5′ . . . GGATCC . . . 3′ 5′ . . . AGATCT . . . 3′ 5′ . . . GGTACC . . . 3′ site 3′ . . . CCTAGG . . . 5′ 3′ . . . TCTAGA . . . 5′ 3′ . . . CCATGG . . . 5′ overhang 5′ . . . G GATCC . . . 3′ 5′ . . . A GATCT . . . 3′ 5′ . . . GGTAC C . . . 3′ 3′ . . . CCTAG G . . . 5′ 3′ . . . TCTAG A . . . 5′ 3′ . . . C CATGG . . . 5′ Table 2.1: Restriction endonuclease digestion sites used in this thesis. These enzymes (NEB, Ipswich, MA) are used in this study to prepare the DNA plasmids. The gene of interest is flanked by BamHI site in the front and BglII, KpnI sites in the back. The BglII and kpnI sites in the back allows for insertion of a second gene of interest in between the two sites. The BglII/BamHI hybrid site formed in this fashion can not be digested by neither BglII nor BamHI. 2.2 Coupling reaction Purified protein samples are concentrated to∼6-8 mg/mL using the Amicon Ultra- 4 Centrifugal filter unit with Ultracel-3 membrane (MILLIPOLE, Billerica, MA) 20 2.3. Instrumentation of AFM and reacted with the cross-linker BM(PEO)3 (1,8-bis-maleimido-(PEO)3; Molec- ular Biosciences, Boulder, CO) as previous described [51]. 2.3 Instrumentation of AFM Design of the AFM rest on one remarkably simple idea: interactions between the probe and the sample generate a force. By measuring this force, molecular level information about the sample can be learned. In order to achieve this goal, a small but highly sensitive probe is required. The relative position of the probe and sample needs to be controlled accurately and precisely such that meaningful information can be extracted from such measurements. Figure 2.1: Schematic diagram of an AFM. Typical setup of an AFM is illustrated in Fig. 2.1. The mechanical probe is a cantilever equipped with a sharp tip (typical radius of curvature of the apex is ∼10-100 nm [52]). The probe is capable of operating in vacuum, air, or liquid. 21 2.3. Instrumentation of AFM The sample is mounted on a piezoelectric actuator which serves as a positioner that moves the sample. In the contact mode of operation, force generated from interactions between the probe and sample will cause a deflection in the cantilever. This deflection is usually measured by a optical beam deflection method [53], in which a position sensitive photodiode detector measures the deflection of a laser beam reflecting off the back of the cantilever. The measured deflection is sent to the detector and control electronics system. Both the sample position data and cantilever deflection data are sent to a computer for analysis and display. The AFM has enjoyed great success in nanometre scale imaging applications and atomic resolution images are achievable [52, 54–56]. Readers of this thesis are referred to the references above for more information on imaging techniques. The rest of this section will focus on features of the AFM that are important to force curve mode of operation. There are four subsections: 1. Tip-cantilever assembly, 2. Piezoelectric actuator, 3. Control system, 4. Isolation and noise reduction. 2.3.1 Tip-cantilever assembly Nowadays, tip-cantilever assembly are typically manufactured from silicon nitride using well-established micro-fabrication techniques from the semi-conductor in- dustry [54]. The back of the cantilever is often coated with gold to improve re- flectivity for use with the optical beam deflection detection method. A scanning electron microscope (SEM) image of a commercially available tip-cantilever as- sembly is showing in Fig. 2.2. Typical cantilevers have length and width in micrometres. This means they are easily seen under optical microscope, making alignment of the laser spot straight- forward. The optical beam deflection detection method is simple yet robust. Res- 22 2.3. Instrumentation of AFM Figure 2.2: Cantilever SEM image. Three AFM cantilevers (B,C,D) have dif- ferent dimensions. Reprinted from [57], c©(2011), with permission from Elsevier. olution limit of AFM employing such detection method is typically set by the thermal noise of the cantilever [54]. One important property of an AFM cantilever, as pointed out by the inventors [1], is its natural resonance frequency. Assuming cantilever deflection can be modeled as a simple harmonic oscillator described by classical mechanics (this is an oversimplification but it makes a point), then its resonance frequency is given by: ω0 = √ ks m0 (2.1) where ω0 is natural resonance frequency in radian/s, ks is force/spring constant of cantilever in N/m or pN/nm, m0 is effective mass of cantilever accounting for its shape. In order to detect a small force, the cantilever needs to be as soft as possible. A 23 2.3. Instrumentation of AFM small value of ks will maximize amplitude of cantilever deflection at any given force. At the same time, the resonance frequency need to be kept sufficiently high to minimize coupling to background noise and decrease response time. The most obvious solution of the above to decrease the size of an AFM cantilever (i.e. m0). This is a nice example demonstrating the importance of advances in micro- fabrication techniques to the development of the AFM [1, 54, 56]. 2.3.2 Piezoelectric actuator The piezoelectric actuator utilizes the piezoelectric effect to achieve small move- ment of the sample with high precision, vertical displacement on the order of Ångströms can be achieved [1]. A piezoelectric material will change its shape under an applied electric field. This effect is exploited to control the length of piezoelectric crystal through an applied voltage. High voltage (∼100 V) is needed to achieve the required range of displacement for experiments, thus a high voltage amplifier is required. Many modern AFMs employ the so called tube scanner type actuator due to its compactness. Careful factory calibration of the piezoelectric actuator is required to achieve a high level of accuracy and precision [54, 55]. Al- though the piezoelectric actuator is able to move small distances accurately, it only covers a limited range (∼µm). For this reason, micrometers are needed for initial placement of the sample to bring it within range of the piezoelectric actuator. 2.3.3 Control system The AFM control system, composed of the actuator control system and feedback system, is arguable the most important part of the experimental setup. As dis- cussed briefly in the previous subsection (2.3.2), the piezoelectric actuator and its control system needs to be calibrated carefully. The feedback system can control the piezoelectric actuator through the actuator control system, this can provide ac- tive compensation [54, 55]. For instance, when the experimentalist would like to maintain the interaction force between the AFM tip and the sample at a constant 24 2.4. Force spectroscopy with AFM level. AFM control system can be quite complex and specialized depending on the application field, further details are beyond the scope of this thesis. 2.3.4 Isolation and noise reduction The AFM is very sensitive to mechanical noise from the environment. Vibration isolation and damping equipment reducing transfer of building vibration is essen- tial. Ideally, the AFM should be fully enclosed by a suitable box to further isolate it from acoustic noise and temperature changes for optimal performance [54, 55]. 2.4 Force spectroscopy with AFM 2.4.1 Cantilever spring constant calibration In order to use the AFM to measure force, sensitivity of the mechanical probe needs to be calibrated [58]. Proper calibration of cantilever string constant is a whole research field in itself [22, 23], only the method used for work in this thesis will be presented for brevity. As outlined in section 2.3, forces generated form molecular level interactions deflect the AFM cantilever from its equilibrium position. The optical beam deflection system used to detect this deflection outputs this deflection signal as a voltage difference between the split photodiode detector. The first step in the calibration involves the determination of the inverse opti- cal lever sensitivity (InvOLS) with the unit of nm/V. This constant will be used to convert raw deflection in unit of V to cantilever deflection (xc) in unit of nm. This is performed taking advantage of the highly precise piezoelectric actuator. A hard substrate surface (e.g. glass) is moved towards the cantilever at a constant velocity and raw deflection in volt is measured. The movement is reverse when a preset trigger in raw deflection is reached. At this point, the substrate is retracted to some initial position away from the cantilever. Plot of piezoelectric actuator dis- placement (xp) versus raw deflection yields a so called approach-retraction curve. Assuming the substrate is non-deformable, cantilever deflection will equal to ac- 25 2.4. Force spectroscopy with AFM tuator displacement (i.e. xc = xp) in the contact/linear region of the curve. Slope of this linear region is the reciprocal of InvOLS. The second step in the calibration involves the determination of the spring constant with the unit of pN/nm. The thermal noise method [59, 60] is popular due to its ease of implementation. If we assume the AFM cantilever behaves as a simple harmonic oscillator, then potential energy stored in the cantilever is 1 2ksx 2 c . Equipartition theorem states that average potential energy of the oscillator in thermal equilibrium is 12kBT . The following equality can be established for the cantilever in thermal equilibrium: 1 2 kBT = 1 2 ks 〈 x2c 〉 , (2.2) where 〈 x2c 〉 denote time averaged square of cantilever deflection driven by thermal fluctuation. This quantity can be measured directly by monitoring the deflection of an AFM cantilever in the time domain after thermal equilibration. Alternatively, it can be accessed in the frequency domain by using Parseval’s theorem which states∫ |xc(t)|2 dt = ∫ |x̄c(ω)|2 dω, (2.3) where x̄c(ω) is the Fourier transform of xc(t). In other words, integral of the square of cantilever fluctuation over the time domain equals integral of the power spectral density over the frequency domain [60]. In the first step, raw deflection in voltage is converted into cantilever deflec- tion (xc) in nm using InvLOS. In the second step, cantilever deflection in nm is converted into force in pN using spring constant of the cantilever. Assuming the cantilever behaves like a Hookean spring, force will be given by F = −ksxc. 2.4.2 Contour length increment The use of polyprotein offers several advantages [62]. Typical constant-velocity AFM experiment is carried out in a fly-fishing fashion. Attachment of molecules to the AFM tip is often non-specific and non-specific interaction is especially 26 2.4. Force spectroscopy with AFM significant when the AFM tip is close to the substrate (within ∼50 nm). The polyprotein can also serve as a linker to extend folded domains further away from surface. Most important advantage with the use of a polyprotein is that it gives a fingerprint for single molecule events. Consider the typical case where all of the individually folded domains are identical. Assuming the folded domain is inextensible, rupture of one folded do- main will add to the effective contour length of the polyprotein. Sequential rup- tures of the rest of folded domains will add the same increments to the contour length. The example in Fig. 2.3 illustrates the idea of the contour length incre- ment. Since the polyprotein in this example has only one type of domain, contour length increments are identical (28.0 ± 0.3 nm, mean ± SD calculated from values in figure). Segment of polypeptide in folded domain is “hidden” to the applied force and is not accounted for in the contour length of the polyprotein. Rupture of one folded domain let the applied force “see” this segment of polypeptide and adds to the contour length of the polyprotein. Contour length of one unfolded domain is estimated to be 0.36 nm/aa × number of aa. The expected contour length incre- ment can be calculated based on knowledge of protein structure. Initial distance between attachment residues can be calculated from an atomic resolution protein structure using a computer program such as VMD [63]. Contour length increment will be given by contour length of the unfolded domain subtracted by the initial distance. The contour length increment can be experimentally determined from a force versus extension plot. Fitting the rising edge of each force peak with force ver- sus extension expression of the WLC (eq. 1.1) will give the contour length of each. Taking the difference between successive fits will yield the contour length increment. Experimental determined value of contour length increment serves as an important fingerprint and reassure the experimentalist that it is truly a single molecule is being mechanically denatured. 27 2.4. Force spectroscopy with AFM Figure 2.3: Contour length increment as single-molecule fingerprint. Panel (a): change in end to end distance is given by Lc − L0, when a folded polypeptide chain is extended to its contour length. Panel (b): a folded domain is essentially inextensible. Unfolding of a folded domain adds to the effective contour length of the polyprotein. Panel (c): successive WLC fits are overlaid on a data trace. Panel (b) and (c) reprinted from [61], c©(1999), with permission from Elsevier. 28 2.4. Force spectroscopy with AFM 2.4.3 Rupture force statistics Due to the stochastic nature of single molecule processes, statistical data on rup- ture force needs to be collected in constant-velocity AFM experiment. Rupture force is defined as the height of an individual force peak as measured from a base- line (see Fig. 2.4). There are eight (8) peaks featured here, but only seven (7) rupture force values can be collected. The last peak must be excluded because it could be a detachment peak. Rupture force statistics needs to be further binned into a histogram and a normalized rupture force histogram is used to estimate the density of a distribution. In other words, a normalized rupture force histogram is described by the probability density function of an appropriate distribution. What is the appropriate probability density function for the normalized rup- ture force histogram obtained from constant-velocity AFM experiments? In such an experiment, extension is increased at a constant rate (x = vt, where v is called pulling velocity). Before the unfolding of any domain, force versus extension of the polyprotein is described by the interpolation formula of WLC. Substitu- tion of x = vt and F(x) (equation 1.1) into equation 1.4 yields the appropriate rate equation. It can be shown that there is no trivial closed-form solution to the rate equation for the unfolding kinetics without making certain assumptions [65]. Instead of making and justifying assumptions, a kinetic type Monte Carlo simula- tion (see [66, 67]) has been devised by others to reproduce the elastically coupled behaviour of a two-state unfolding model [25, 68–70]. 2.4.4 Monte Carlo simulation procedure As introduced in the last subsection, the Monte Carlo scheme is devised to re- produce events in a constant-velocity AFM experiment [68]. In this scheme, the polyprotein is being extended by the AFM tip at a constant velocity v starting from zero extension (x = 0, t = 0). The polyprotein have N f folded domains each with contour length I f and Nu unfolded domains each with contour length Iu. Contour length of the polyprotein L is given by L = N f I f + NuIu. 29 2.4. Force spectroscopy with AFM Figure 2.4: Information from rupture force statistics. Panel A: rupture force is measured from the baseline (dashed horizontal line) to top of each peak. Blue lines are fits using equation 1.1 (subsection 1.2.2). Panel B: experimentally col- lected rupture force values are binned into a histogram. Red line is generated from a Monte Carlo simulation procedure (see subsection 2.4.4 for more on the proce- dure). Panel C: average rupture force is plotted as a function of pulling velocity. Black line is generated from a Monte Carlo simulation procedure. Reprinted from [64], c©(2000), with permission from Elsevier. At fist, a discrete time step (∆t) is taken to advance the system with corre- sponding extension ∆x = v∆t. At this point, ∆t is added to t and ∆x is added to x. After the time step, the force experienced by the polyprotein is given by equation 1.1. The unfolding rate constant is given by 1.3. The probability of observing an unfolding event is Pu = N f ku(F)∆t. This probability is compared to a random number1 R drawn from a continuous uniform distribution on (0, 1). If Pu > R, one 1Actually, a deterministic machine like a personal computer is only capable of generating pseudo-random numbers. The philosophies and algorithms are beyond the scope of this thesis. Interested readers are kindly referred to [71, 72] and references therein. 30 2.5. SMD simulations domain unfolds in the polyprotein, else the scheme goes back to the first step for another cycle. If one domain unfolds in the polyprotein, then the force F at time t is recorded as the rupture force and N f ,Nu are incremented accordingly before going back to the first step. The scheme is repeated until all of the domains in the polyprotein unfold. If the Monte Carlo scheme is repeated, then rupture force statistics can be collected on these simulations. Recall that equation 1.3 is used to calculate the unfolding rate constant. In that equation, intrinsic unfolding rate k◦u and unfolding distance xu are two parameters which need to be set. Rupture force statistics from simulation is made to match that from experiments, as judged by visual inspection (see Fig. 2.4), by manually adjusting these two parameters. 2.5 SMD simulations Rapid increase in computational power has fueled the development of physics based simulations of macromolecules [73–75]. Molecular dynamics simulation can be used to monitor the temporal evolution of a biological macromolecule by numerically solving Newton’s equation for the classical many-body problem. In steered molecular dynamics (SMD), an external applied force (hence “steered”) is used to perturb the system. Atomistic information can be extracted from the simulated trajectories [26, 76]. All-atom SMD simulations have had great success in explaining experimental results from single-molecule AFM and have made predictions which were later confirmed by experiments [42, 43, 76–78]. It has become an invaluable method complimentary to experiments, and are routinely combined with single-molecule AFM to reveal more insights [79–83]. For more detailed overviews of the ap- plication of SMD to mechanical denaturation of proteins, the readers are kindly referred to the references [84, 85]. 31 Chapter 3 Molecular Mechanism Underlying Mechanical Anisotropy Mechanical responses of elastic proteins are crucial for their biological function and nanotechnological use. Loading direction has been identified as one key de- terminant for the mechanical responses of proteins. However, it is not clear how a change in pulling direction changes the mechanical unfolding mechanism of the protein. Here, we combine protein engineering, single-molecule force spec- troscopy, and steered molecular dynamics simulations to systematically investi- gate the mechanical response of a small globular protein GB1. Force versus ex- tension profiles from both experiments and simulations reveal marked mechanical anisotropy of GB1. Using native contact analysis, we relate the mechanically ro- bust shearing geometry with concurrent rupture of native contacts. This clearly contrasts the sequential rupture observed in simulations for the mechanically la- bile peeling geometry. Moreover, we identify multiple distinct mechanical un- folding pathways in two loading directions. Implications of such diverse unfold- ing mechanisms are discussed. Our results may also provide some insights for designing elastomeric proteins with tailored mechanical properties. 2 2A version of this chapter has been published [86]: Li, Y. D., Lamour, G., Gsponer, J., Zheng, P. & Li, H. B.The Molecular Mechanism Underlying Mechanical Anisotropy of the protein GB1. Biophys. J. 103, 2361-2368 (2012) c©Elsevier. The abstract, excerpts and parts are reproduced here with permission. 32 3.1. Introduction 3.1 Introduction Mechanical strength is an important trait for many proteins. It is crucial not only for their functions in biology [17], but also for their possible utilization in novel protein-based material with tailored mechanical properties [9]. Single- molecule force spectroscopy [25, 39, 40] and steered molecular dynamics (SMD) simulations [42–44] have greatly expanded our knowledge of the molecular ori- gins for mechanical strength. Previous studies have demonstrated native topolo- gies [24, 43, 44] and detailed interactions, such as hydrophobic packing in pro- tein structures [87, 88], play pivotal roles in the determination of mechanical properties for a protein [6, 8]. Furthermore, mechanical response of proteins to a stretching force is anisotropic and depends strongly on the loading direction [46, 77, 78]. Two early single-molecule atomic force microscopy (AFM) studies [77, 78] have shown mechanical strength of globular domains depends strongly on loading direction. A mechanically strong protein can be surprisingly compli- ant when stretched along an appropriate mechanically weak direction. In single-molecule force spectroscopy, the loading direction is determined by the attachment points in an oligomeric protein polymer, several strategies have been develop to control the linkage chemistry [45, 46, 51, 89]. Using polypro- teins (i.e., a linear oligomeric protein made by covalently linking monomeric do- mains) produced using these methods, researchers have shown that an individual protein molecule exhibits fascinating anisotropic mechanical responses to stress from different loading direction [46, 80]. Recently, Lee and colleagues have ex- plored the mechanical anisotropy of ankyrin repeats by exploiting order of domain placements in the polyprotein molecule [83]. Combining single-molecule AFM experiments and coarse-grained SMD simulations, these researchers have attribute this anisotropy to the different ways in which native contacts in ankyrin repeats are broken between neighbouring α-helices. Despite of these efforts, it is still unclear how different loading directions, achieved via changes in anchoring residues, alter the mechanical unfolding mech- anism of a mechanically strong protein. To address this question, we attempt to 33 3.2. Experimental section explore possible anisotropy in the mechanical response of a small globular pro- tein under an applied mechanical perturbation. GB1, the B1 immunoglobulin G binding domain of streptococcal protein G, is chosen as the model protein in this study due to its small size and well documented mechanical properties [28, 90]. Mechanical unfolding mechanism of this protein (56 aa) from its two ends has been identified to be different from that of chemical denaturant unfolding, at the atomic level based on SMD simulation results [79, 81, 82]. Recently, Graham and Best have explored the switch of unfolding pathways under a pulling force with the use of a coarse-grained Gō-like model [91]. They have discovered the switch from the intrinsic unfolding pathway to the novel mechanical pathway can be either abrupt or gradual, depending on the loading direction. Protein engineering, single-molecule AFM experiments, and all-atom SMD simulations are combined in this study to systematically investigate the possible anisotropic mechanical response of GB1. Results from this current study supports findings of Graham and best [91], in which significant anisotropy is observed in the mechanical resoponse of GB1. Insights on mechanical unfolding mechanisms are derived from all-atom SMD simulations. The origin of such anisotropy is discussed on the basis of protein structure and mechanical unfolding mechanism. 3.2 Experimental section 3.2.1 Protein engineering Gene encoding GB1 was a generously gift from Dr. David baker at the Univer- sity of Washington. Cysteine mutations on gene encoding GB1 were performed using a standard site-directed mutagenesis method. Gene encoding mutant con- structs were subcloned into the expression vector pQE80L (QIAGEN, Valencia, CA). All constructs were overexpressed in Escherichia coli strain DH5α and pu- rified using Co2+ affinity chromatography with TALON His-Tag purification resin (Clontech Laboratories, Mountain View, CA). The purified protein samples (∼2 34 3.2. Experimental section mg/mL) were kept in phosphate-buffered saline (PBS, 300 mM NaCl, pH 7.4) at 4◦C. The PBS contains 150 mM imidazole, which was required in the purification procedure. Names of the mutant constructs contain two numbers, which are the aa index numbers for the two engineered cysteine residues (e.g. G10-48C, residues 10 and 48 are mutated into cysteine residues). Letter G stands for GB1 and letter C stands for cysteine. 3.2.2 Thiol-maleimide coupling reaction In a typical experiment, purified protein samples were concentrated to ∼6 mg/mL. Concentration of protein sample was carried out with the Amicon Ultra-4 cen- trifugal filter unit equipped with Ultracel-3 membrane (MILLIPORE, Billerica, MA). The concentrated protein sample was reacted with the chemical cross-linker BM(PEO)3 (1,8-bis-maleimido-(PEO)3; Molecular Biosciences, Boulder, CO) as previously described [51]. The reaction mixture was incubated at 37◦C for ∼8 hours and stored at 4◦C. Aliquots of the cross-linked protein samples were used directly in AFM experiments. 3.2.3 Single-molecule AFM experiments Single-molecule AFM experiments were carried out on a custom-built AFM. The construction of this AFM was described previously [92]. Spring constants of the Silicon Nitride MLCT cantilevers (Bruker, Santa Barbara, CA) were calibrated in PBS before each experiment using the thermal noise method [60]. Typical value for the spring constant is ∼50 pN/nm. In a typical experiment, ∼1 µL of cross-linked protein sample was deposited onto a clean glass coverslip covered with ∼50 µL PBS. The sample was allowed to adsorb for ∼5 minutes. Constant- velocity AFM pulling experiments were performed at 400 nm/s unless otherwise noted. Contour length increment was calculated by subtracting the initial distance between the two tethered Cα atoms from the estimated distance between the two 35 3.2. Experimental section in the fully stretched protein. The estimated distance was calculated using the formula 0.36 nm/aa × number of aa between the two Cα atoms. Graham and Best [91] chose to place one tethered Cα atom on the α-helix (residue number 32). This would generally result in contour length increments much shorter than 10 nm, regardless of where the other tethered Cα atom was placed (residue 10 or 56). Relatively short polyprotein would create practical obstacles in AFM pulling experiments. Even though the spatial resolution of AFM in the vertical direction is on the Å level, non-specific interactions are hard to avoid at extension < 50 nm. 3.2.4 SMD simulations GB1, starting from the initial PDB structure 1PGA, was equilibrated for 1 ns during which it is reasonably stable, with Cα root mean-square deviation in the range of 2 Å. The equilibrated final structure was used as the starting point in constant velocity (0.1 Å/s) SMD pulling simulation performed with the AFM module of CHARMM [73, 74]. Each protein construct was simulated by tethering Cα atoms at two amino acid positions corresponding to the cysteine mutations. The protein was then subjected to constant velocity stretching between the two tethered Cα atoms, in the direction parallel to the line connecting the two at the beginning of the simulation. All simulations were carried out at 300 K in the CHARMM22 [93] force field with CMAP correction [94] using the implicit solvent FACTS [95]. The choice of the implicit solvent over explicit solvent was based on the consideration of computational efficiency, as we wanted to carry out all-atom simulations (param22 parameter file) and a rather large number of trajectories needed to be collected. Calculations used an atom-based truncation scheme with a list cutoff of 14 Å, a nonbond cutoff of 12 Å, and the Lennard-Jones smoothing function initiated at 10 Å. Electrostatic interactions were force shifted. The SHAKE algorithm [96] was used for covalent bonds involving hydrogen atoms enabling integration time steps of 2 fs. All native contact analyses were performed under VMD [63] using the root 36 3.3. Design of loading directions mean-square deviation trajectory tool enhanced with a native contact plug-in. Two atoms were considered in contact if their centres are within a distance of 5 Å and only native contacts between heavy atoms were considered. For backbone atom- atom contacts, only Nitrogen and Oxygen atoms from residues that are separated by at least two residues were considered. For side-chain atom-atom contacts, contacts between two adjacent residues and within one residue were not taken into account. 3.3 Design of loading directions It is well known that the shear topology, referring to the unfolding force being applied close to parallel to neighbouring β-strands is a critical criterion in deter- mining the mechancial strength of proteins. This puts restrictions not only on protein structure (i.e. native topology) but also on the pulling axis (i.e. loading direction). In the context of this current study, a pulling axis is defined by two aa residues of the protein domain which serve as the anchors in mechanical unfold- ing. To design the pulling axes for probing mechanical anisotropy, we first look at the three dimensional structure of GB1. Structure of GB1 has been solved at atomic resolution with both NMR spectroscopy [97] and X-ray crystallography [98]. Overall tertiary structure of GB1 consists of a four stranded β-sheet packed against a long α-helix. The folding topology of GB1 belongs to the so called β-grasp [99] or UB-roll [100] folding motif. The β-sheet can be further broken down into two smaller structural elements, namely the N-terminal and C-terminal β-hairpins (see Fig. 3.1). shearing geometry enforced by the arrangement of termi- nal strands constitutes the main point of mechanical resistance [24, 79]. Graham and Best [91] have identified two classes of pulling axes based on force-dependent unfolding kinetics from simulation on GB1 using a coarse-grained Gō-like model. The mechanically strong class has their pulling axes aligned approximately along the long axis of the β-sheet; the mechanically weak class has their pulling axes lie approximately in between the β-sheet and α-helix. We follow similar rationales in 37 3.3. Design of loading directions the design of pulling axes (mechanically strong: G10-48C, G1-40C, and G19-56C where the numbers are indices of achoring residues). However, it is challenging to work with the mechanically weak class of pulling axes identified by Graham and Best [91] due to technical caveats discussed in subsection 3.2.3. Instead of axes lying between the β-sheet and the α-helix, we choose pulling axes roughly perpendicular to the long axis of the β-sheet (Fig. 3.1, G19-48C and G10-40C). Figure 3.1: Ribbon cartoon representation of the structure of GB1. Cartoons based on PDB structure 1PGA are rendered with VMD [63]. The β-strands are numbered from N-terminus to C-terminus. In a previous study, researchers have developed a protocol for constructing polyproteins with linkages between two precisely controlled amino acid residue 38 3.4. Anisotropic mechanical response of GB1 positions based on thiol-maleimide coupling chemistry [51]. This method is sim- ilar to the disulphide bond method developed by Rief and co-workers [46], which has been used to construct polyproteins. We want to emphasize that the load- ing direction in AFM experiments is defined by the linkages in the polyprotein. In this approach, we need to mutate two native residues in the protein into cys- teine residues. Because native structure of GB1 lacks cysteine residues, concerns about formation of unwanted linkages are eliminated. Moreover, these two na- tive residues need to be solvent accessible and sufficiently far apart to avoid in- tramolecular linkage formation. Based on these criteria, a total of five bi-cysteine constructs have been made (see Fig. 3.1). 3.4 Anisotropic mechanical response of GB1 To investigate the effects of loading direction on the mechanical unfolding of GB1, we have carried out constant-velocity single-molecule force spectroscopy experiments on each of the bi-cysteine GB1 constructs. The force-extension trace obtained from stretching a polyprotein of a construct displays a characteristic saw- tooth pattern (e.g., see bottom trace in Fig. 3.2 A). Each individual force peak in the sawtooth pattern is the result of mechanical unfolding of each individual do- main in the polyprotein chain, except for the last peak, which corresponds to the extension of the unfolded polyprotein and subsequent detachment from either the AFM tip or glass substrate. Here, we want to emphasize that mechanical unfold- ing of each domain happens between the engineered cysteine residues. The rising edge of each force peak is well described by the worm-like chain model of poly- mer elasticity [101]. Contour length increments ∆Lc, calculated from a successive worm-like chain fit to the force peaks, agree well with expected values from struc- tural considerations (see subsection 3.2.3). These two observations taken together suggest the folded protein domains are completely unraveled between the points of attachment in an all-or-none fashion. 39 3.4. A nisotropic m echanicalresponse ofG B 1 Figure 3.2: Anisotropic response of GB1 to mechanical stress. A version is published in [86], c©(2012) Elsevier. 40 3.4. Anisotropic mechanical response of GB1 Typical force versus extension traces from constant-velocity single-molecule AFM experiments are presented in Fig. 3.2 panel A. Dashed lines are fits using the interpolation force versus extension formula of the WLC model (see equation 1.1 and [101]). Typical persistence length p used is ∼0.4 nm. Normalized frequency histograms of unfolding forces are presented in Fig. 3.2 panel B. Gaussian fits plotted as solid curves are shown to guide the eye. Cartoons similar to those shown in Fig. 3.1 are displayed for easy comparison. Because mechanical unfolding of a protein is stochastic, unfolding force of a protein will fluctuate randomly around a mean value. The unfolding force his- tograms of the five constructs, at 400 nm/s pulling velocity, are reported in Fig. 3.2 B. G1-40C unfolds at a mean force of ∼110 pN, making it the mechanically strongest construct of the five constructs studied, whereas the mechancially weak- est G10-40C construct unfolds at ∼40 pN. We note that even the strongest con- struct unfolds at significantly lower force than wt GB1, which unfolds at ∼180 pN when it is stretched from its N-C termini [90], suggesting that loading direction along the N-C termini remains the most mechanically resistant geometry. From the rupture force histograms, it is clear that the distributions of rupture forces also depend strongly on the loading direction. Furthermore, we observe that both the mean and variance of rupture force values are larger when the loading vector is close to parallel to the long axis of the β-sheet. This is the case for constructs G1- 40C, G19-56C, and G10-48C. On the other hand, both values are smaller when the pulling vector is not aligned with the long axis of the β-sheet as seen in the histograms for constructs G10-40C and G19-48C. To further characterize the mechanical unfolding of GB1 under different load- ing geometries, we examine dependency of mean rupture force on the pulling velocity by measuring the force-extension relationships under different pulling ve- locities. Because mechanical unfolding of all constructs happens in an all-or-none fashion, the kinetics of their mechanical unfolding can be modeled as a two-state system with a force-dependent unfolding rate constant. The Bell model [34] char- acterizes mechanical unfolding kinetics using the unfolding rate constant at zero 41 3.4. Anisotropic mechanical response of GB1 Figure 3.3: Dependency of unfolding force on the pulling velocity. Kinetic parameters can be estimated from this (see 2.4.4 [25, 68, 70]). Experimentally determined unfolding forces are represented as symbols with error bars (mean ± SD). Reprinted from [86], c©(2012), with permission from Elsevier. force k◦u and the unfolding distance xu. These two parameters can be extracted from pulling velocity dependency data (Fig. 3.3) using Monte Carlo simulations based on a published protocol [70]. From the values tabulated in Table 3.1, we note that pulling velocity depen- dency of the unfolding force of G19-48C and G10-48C can be adequately re- produced using the same rate constant k◦u as wt GB1 but with different values of unfolding distance xu. However, the k◦u for GB1 fails to reproduce the behaviours of G19-56C and G1-40C. For these constructs, both k◦u and xu have different val- ues when compared to wt GB1. According to the values of xu, the constructs can 42 3.4. Anisotropic mechanical response of GB1 Construct Fu (pN) xu (nm) k◦u ( s−1 ) GB1 178 ± 40 0.17 0.039 G10-40C 36 ± 9 – – G19-48C 52 ± 16 0.58 0.039 G10-48C 92 ± 19 0.33 0.039 G19-56C 86 ± 26 0.31 0.089 G1-40C 109 ± 23 0.38 0.0065 Table 3.1: Kinetic data on mechanical unfolding of GB1 constructs. These data are derived from constant-velocity single-molecule AFM experiments and Monte Carlo simulations. Fu denotes unfolding force and is reported as mean ± SD (pulling velocity is 400 nm/s). xu denotes the unfolding distance, which is the distance between the native state and transition state of mechanical unfolding. k◦u denotes the unfolding rate constant at zero force. Pulling velocity dependency for G10-40C has not been examined due to low Fu close to the detection limit of our AFM set-up (∼20 pN). Values for GB1 are taken from previously published results [90]. Typically Monte Carlo simulations yield xu values accurate to within ±0.5 nm and k◦u values within three-fold. be divided into three groups with small (GB1), intermediate (G1-40C, G19-56C, and G10-48C), and large (G19-48C and possibly G10-40C) xu values. A smaller xu value implies the transition state of mechanical unfolding is highly native-like. Variations in the mean and variance of the rupture force reflect differences in the underlying mechanical unfolding free energy profiles. The mean rupture force is determined by both k◦u and xu. Because barrier crossing in mechanical unfolding is thermal driven (hence stochastic), the variance/standard deviation of the unfolding forces is governed by the relative magnitude of the unfolding distance compared to thermal energy (kBT/xu). Therefore, differences in unfolding kinetics/mechanism will lead to changes in the mean and variance/standard deviation of the unfold- ing force distribution. It is reasonable to anticipate that unfolding in mechanical strong (high mean value of rupture forces) and weak (low mean value) directions proceed via different molecular mechanisms. It is of note that visual inspections on the five constructs (Fig. 3.1) reveal that shearing geometry is roughly main- 43 3.5. Mechanical unfolding mechanisms tained in the cases of G10-48C, G19-56C, and G1-40C; whereas G10-40C and G19-48C are arranged to unfold the protein in a peeling or un-zipping fashion. 3.5 Mechanical unfolding mechanisms To learn about the mechanical unfolding mechanism for each construct, we turn to SMD simulations. Results from SMD studies have been shown to correlate well with single-molecule AFM results [42, 43, 76]. Previous results [79, 81, 82, 102] have indicated that terminal β-strands 1 and 4 of GB1 (Fig. 3.1) are in direct con- tact and form a mechanical clamp motif that resists mechanical stress and protects the protein from unfolding. Previous SMD simulations on GB1 have shown the mechanical unfolding, between its N-C termini, initiates with separation of the terminal β-strands 1 and 4 [79, 81, 82]. We first carried out similar SMD simulations on GB1, which shall serve as the benchmark for latter comparisons. One representative force-extension plot from such a simulation is shown in Fig. 3.4 F. The force peak at small extension (<10 Å) is the main event in the graph, which corresponds to the burst of the mechan- ical clamp. Fig. 3.4 F also shows a snapshot of GB1 taken right after the burst from the simulation trajectory. The snapshot shows that terminal β-strands 1 and 4 are separated from each other. At the same time, the C-terminal β-hairpin is separated from the α-helix and the N-terminal β-hairpin. In some simulation tra- jectories (for example, see Fig. 3.4 D1), there may also be one or two minor force peaks, corresponding to mechanical unfolding intermediates not seen in experi- ments, following the main force peak. All of the notable features in the simula- tions on GB1 agree well with results from previous studies [79, 81, 82]. Two of these previous studies have been carried out with explicit solvent model while one study is carried out with implicit solvent model. Such close agreements between simulation results suggest mechanical unfolding pathways identified from SMD simulations are robust and relatively insensitive toward differences between the chosen solvent models. 44 3.5. Mechanical unfolding mechanisms Figure 3.4: Force versus extension plots with snapshots. Traces in lighter colours are smoothed from raw traces in darker colours with a moving median (box size: 21 points). Ribbon cartoon snapshots illustrate protein structures are taken with VMD just after main burst events. A version is published in [86], c©(2012) Elsevier. 45 3.5. Mechanical unfolding mechanisms To directly compare between different loading directions and elucidate the ef- fects of loading directions on the mechanical unfolding mechanism, we carried out constant-velocity SMD simulations for all of the protein constructs. Typical force versus extension plots are shown in Fig. 3.4 along with one snapshot from each simulation trajectory. For protein constructs G1-40C and G19-56C, we have included two plots from each because there are two apparent unfolding pathways. Even though rupture force values obtained from SMD simulations cannot be di- rectly compared to experimental values due to the vast difference in timescale of the two approaches, we note that our simulation results correctly predict the rel- ative rank of mechanical robustness. Indeed, our SMD simulations predict that the unfolding force values will decrease from ∼1.5 nN (for GB1) to ∼1.2 nN (for G10-48C,G1-40C, and G19-56C) and finally to non-detectable (for G10-40C and G19-48C), in excellent agreement with experimental results. Comparing Fig. 3.4 F and C, it appears that G10-48C unfolds through a path- way very similar to that of wt GB1. However, force versus extension plots for G1-40C and G19-56C are much more complex. In their previous study com- bining all-atom and coarse-grained simulations [103], West and colleagues have found that native interactions are more important that non-native contacts in deter- mining the origin of mechanical strength. Here, we monitor the fractions of native contacts between and within secondary structural elements (namely: N-terminal β-hairpin, C-terminal β-hairpin, and α-helix) for each simulation trajectory. Back- bone native contacts (i.e., backbone hydrogen bonds) within structural elements are used as a gauge for the loss of secondary structure; whereas side-chain na- tive contacts between structural elements as well as backbone hydrogen bonds between the β-hairpins (i.e., formed between terminal β-strands 1 and 4) are used to track the loss of tertiary structure. We have performed this analysis on all of the protein constructs and the results are shown in Fig. 3.5. 46 3.6. Detailed mechanisms from native contact analysis 3.6 Detailed mechanisms from native contact analysis In the case of wt GB1, native contact analysis reveals significant loss of native contact between C-terminal β-hairpin and α-helix during burst of the mechani- cal clamp at an extension of ∼5 Å. Burst of the mechanical clamp not only in- cludes rupture of backbone hydrogen bonds between β-hairpins (Fig. 3.5 F4), but also includes substantial loss of native contacts between the side chains (see Fig. 3.5 F1, F2, F3). Following the main burst event, there are also some disrup- tions to the backbone hydrogen bonds within each structural element especially the C-terminal β-hairpin. The overall picture of the partially unfolded GB1 from our SMD simulation involves rupture of backbone hydrogen bonds between β- hairpins and detachment of the C-terminal β-hairpin from the core. This is in close agreement with previous results [79, 81, 82]. Native contact analysis, on simulation trajectories for G10-48C (Fig. 3.5 C1, C2, C3, C4), shows the mechanical unfolding pathway is very similar to that of wt GB1. The slight differences may account for its lower mechanical stabil- ity. Mechanical unfolding of G10-48C is initiated with the concurrent rupture of backbone hydrogen bonds between β-hairpins just like that of wt GB1. How- ever, both β-hairpins lose native contacts with the α-helix at roughly the same rate, which clearly contrast the asymmetry in the unfolding of wt GB1. There are some disruptions to the backbone hydrogen bonds in the C-terminal β-hairpin, but these disruptions quickly diminish after the main burst event. The fact that the C-terminal β-hairpin is not directly subjected to mechanical stress in G10-48C, unlike the case for wt GB1, may explain this behaviour. Mechanical unfolding of wt GB1 is a much more cooperative process which generates disruptions to almost all parts of the protein, when compared to that of G10-48C. Lower me- chanical strength of G10-48C, compared to wt GB1, is also in accordance with a conclusion drawn in our previous study where neighbouring β-strands were found to provide critical stabilization [104]. 47 3.6. D etailed m echanism s from native contactanalysis Figure 3.5: Fraction of native contacts between structural elements. Reprinted from [86], c©(2012), with permission from Elsevier. 48 3.6. Detailed mechanisms from native contact analysis G1-40C and G19-56C are the other two constructs found with shearing geome- try. SMD simulations on these constructs show two distinct mechanical unfolding pathways for each construct. In the case of G1-40C, both pathways differ with that of wt GB1 in the sense that the N-terminal β-hairpin is the one detached from the core. In one pathway (Fig. 3.4 E1), concurrent rupture of hydrogen bonds be- tween β-hairpins is accompanied by sequential rupture of hydrogen bonds within the N-terminal β-hairpin and the main burst event. In the other pathway (Fig. 3.4 E2), hydrogen bonds within the C-terminal β-hairpin are ruptured concurrently along with sequential rupture of hydrogen bonds between β-hairpins. Main burst events for both pathways occur at a larger extension value than those of wt GB1 and G10-48C. In the case of G19-56C, one mechanical unfolding pathway (Fig. 3.4 D2) highly resembles that of wt GB1. The other pathway (Fig. 3.4 D1) in- volves the concurrent rupture of hydrogen bonds within the N-terminal β-hairpin accompanied by sequential rupture of hydrogen bonds between β-hairpins. G10-40C and G19-48C are the two constructs found with peeling geometry. SMD simulations show these constructs unfold via similar pathways that involve the sequential rupture of backbone hydrogen bonds. In the case of G19-48C (Fig. 3.4 B), the C-terminal β-hairpin detaches from the core like in the case of wt GB1. However, the peeling geometry causes the backbone hydrogen bonds between β- hairpins to rupture sequentially rather than concurrently. In the case of G10-40C, sequential rupture of backbone hydrogen bonds happen within the C-terminal β- hairpin (Fig. 3.4 A). The N-terminal β-hairpin detaches from the core like in the case of G1-40C. In Fig. 3.5, gray traces are derived from 10 individual SMD simulations and the black trace is arithmetic average of them. Columns one to seven contain the fraction of intact native contacts between various structural elements during me- chanical unfolding (from left to right): side-chain contacts between N-C terminal β-hairpins (N+C hairpin), side-chain contacts between N-terminal β-hairpin and α-helix (N h.p.+helix), side-chain contacts between C-terminal β-hairpin and α- helix (C h.p.+helix), backbone contacts between N-C terminal β-hairpins (N+C 49 3.7. Diverse unfolding mechanisms lead to anisotropy hairpin), backbone contacts within the N terminal β-hairpin (N hairpin), backbone contacts within the C terminal β-hairpin (C hairpin), and backbone contacts within the α-helix (helix). 3.7 Diverse unfolding mechanisms lead to anisotropy In this study, we have combined single-molecule AFM and SMD simulations to explore the anisotropic response of a small globular protein to mechanical stress. Single-molecule AFM experiments have revealed marked directional anisotropy in the mechanical response, whereas SMD simulations have clarified its origin based on unfolding mechanisms. To the best of our knowledge, this is the first time that mechanical anisotropy of a protein has been systematically explained in detailed mechanistic terms based on all-atom SMD simulations. Topological differences between shearing (G10-48C) and peeling (G19-48C) geometries have been linked with mechanistic differences between concurrent and sequential rup- tures of critical backbone hydrogen bonds. This finding using only one protein is in line with previous findings using two proteins with different geometries [24]. Perhaps the most interesting finding from the SMD simulations is the presence of parallel unfolding pathways. Simulations reveal both G1-40C and G19-56C have two apparently distinct unfolding pathways, whereas G10-48C has only one pathway. However, these three constructs exhibit similar behaviours in our exper- iments (in terms of mean and variance in rupture force distribution). Because the simulations are carried out on a timescale that is much faster (∼106 times) com- pared to experiments, these possible parallel pathways must be met with caution. A strategy involving the redesign of unfolding pathways, which has been used in our previous study [105], offers one method to test these predictions. In such a scenario (e.g., take the case of G19-56C), a disulphide bond (or a bihistidine metal binding site [47]) could be engineered at an appropriate site across β-strands 1 and 2 to stabilize the N-terminal β-hairpin. This would result in the wt GB1-like un- 50 3.7. Diverse unfolding mechanisms lead to anisotropy folding pathway (Fig. 3.4 D2) to be favoured and it could possibly shift the mean unfolding force towards a higher value. Recent work by Graham and Best [91] is also concerned with the role of force on the unfolding pathways of GB1. In their coarse-grained Gō-like model based study, Graham and Best have focused on the switch from an intrinsic unfolding pathway to a novel mechanical unfolding pathway [91]. The authors predict a non- linear relationship between the mean rupture force and logarithm of the pulling velocity for wt GB1 and G10-48C type constructs at very low pulling velocity. This phenomenon has not been observed in our experiments, possibly due to rela- tively high pulling velocities used. Large differences between intrinsic unfolding rate constants determined by mechanical and chemical methods [90] suggest that mechanical and chemical unfolding of GB1 follow different unfolding pathways, a result consistent with our recent mechanical ψ-value analysis on the mechan- ical unfolding of GB1 [106]. It would be interesting to observe whether these two alternative pathways do switch at low forces or pulling velocities. To do so, instrumental drift needs to be minimized and force detection limits needs to be improved at the same time. In our present study, we have found that the small globular protein GB1 ex- hibits clear direction anisotropy under mechanical stress. The mechanically most robust construct is the wt GB1 that unfolds at ∼180 pN, whereas the most labile construct G10-40C unfolds at only ∼40 pN. Differences in unfolding mechanisms have been identified to dictate differences in mechanical strength. Because molec- ular determinants of mechanical strength are still not completely understood [8], the rational design of mechanical proteins and materials will be undoubtedly aided by detailed characterization of the unfolding pathways and effects of force on such pathways. 51 Chapter 4 Mechanical Unfolding of an Iron-Sulphur Protein Three dimensional shape of a protein is often crucial to its biological function. Sometimes, folding into a particular shape is not enough and a protein needs the “help” from cofactors such as metal ions and/or small organic molecule. It has been said >30% of proteins in a living cell require cofactors of some sort for their function [50, 107]. The inclusion of a cofactor in the protein folding picture results in further complication to an already complex problem. A natural question arises concerning the sequence of events during the folding and cofactor binding of a protein. Does folding occur before cofactor binding or vise versa? Traditional ensemble based methods have been used to address this problem with much success [107, 108]. As introduced in the introduction section, engi- neered metal binding protein based on GB1 has become a good model system for studying the role of cofactor on protein structure and folding. Recently, Cao and Li have used single molecule AFM to study the folding and binding problem on this model system [50]. They have demonstrated that the binding-after-folding pathway is dominate in this system and a minor folding-after-binding pathway has been identified. This pioneering study shows single-molecule AFM is a promising tool in the study of folding in the presence of cofactor. 52 4.1. Experimental section 4.1 Experimental section 4.1.1 Protein engineering Gene encoding GL5 was contructed previously [109]. Gene encoding wt FdS ([2Fe2S] ferredoxin of type I from spinach) was purchased (GeneScript, Pis- cataway, NJ). Cysteine mutations on gene encoding FdS were performed using a standard site-directed mutagenesis method. Gene encoding mutant constructs were subcloned into the expression vector pQE80L (QIAGEN, Valencia, CA). All constructs were overexpressed in Escherichia coli strain DH5α and purified using Co2+ affinity chromatography with TALON His-Tag purification resin (Clontech Laboratories, Mountain View, CA). All buffer solutions and fractions collected in the purification process were kept on ice whenever possible. The purified pro- tein samples (∼2 mg/mL) were kept in the elution buffer (50 mM Tris-HCl, 1 M NaCl, pH 7.4) at 4◦C. The elution buffer contains 200 mM imidazole, which was required in the purification procedure. UV-visible spectra were taken with a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE). FdS used in most biochemical studies were purified in a step-wise manner from either spinach [110–112] or a recombinant source [113]. The initial purifi- cation sequence was usually a combination of: ammonium sulphate fractionation, gel filtration, and anion-exchange chromatography. The final purification step was usually hydrophobic-interaction chromatography, which helped with the separa- tion of native and denatured ferredoxins [112]. Affinity chromatography is used in our study for simplicity. Judging from the UV-visible absorption spectra, it was found the proteins obtained from such preparations were similar to those obtained after the initial purification sequence of other studies ([113] before hydrophobic- interaction chromatography). 53 4.2. Structure of a plant-type [2Fe2S] ferredoxin 4.1.2 Thiol-maleimide coupling reaction In a typical experiment, purified protein samples were concentrated to ∼8 mg/mL. Concentration of protein sample was carried out with the Amicon Ultra-4 centrifu- gal filter unit equipped with Ultracel-3 membrane (MILLIPORE, Billerica, MA). Concentration of Tris-HCl was increased to ∼100 mM and concentration of imida- zole was reduced to <10 mM. The concentrated protein sample was then reacted with the chemical cross-linker BM(PEO)3 (1,8-bis-maleimido-(PEO)3; Molecu- lar Biosciences, Boulder, CO) as previously described [51]. The reaction mixture was incubated at room temperature for ∼6 hours and divided into ∼5 µL aliquots which were stored at −80◦C. Aliquots of the cross-linked protein samples were used directly in AFM experiments. 4.1.3 Single-molecule AFM experiments Single-molecule AFM experiments were carried out on a MFP3D AFM (Asylum Research, Santa Barbara, CA). Spring constants of the Silicon Nitride MLCT can- tilevers (Bruker, Santa Barbara, CA) were calibrated in buffer before each exper- iment using the thermal noise method [60]. Typical value for the spring constant is ∼45 pN/nm. In a typical experiment, ∼1 µL of cross-linked protein sample was deposited onto a clean glass coverslip covered with ∼100 µL buffer (100 mM Tris HCl, 1 M NaCl, pH 7.4). The sample was allowed to adsorb for ∼10 minutes and ∼5 mL of buffer was added. Constant-velocity AFM pulling experiments were performed at 400 nm/s unless otherwise noted. 4.2 Structure of a plant-type [2Fe2S] ferredoxin Metalloprotein is a large class of protein in which metal ions are used as cofactors in one form or another. Iron-sulphur protein is a big sub-class in the metallo- protein family. Zheng and colleagues have done extensive studies on rubredoxin (protein containing one iron that can change oxidation state) [114, 115]. They 54 4.2. Structure of a plant-type [2Fe2S] ferredoxin have demonstrated that mechanical strength of the protein arises from interaction between the convalently bound iron centre and the protein scaffold. Figure 4.1: Representation of the structure of FdS with iron-sulphur cluster. Mature [2Fe2S] ferredoxin of type I from spinach (FdS for brevity) has 97 aa. Protein part of FdS is depicted in ribbon cartoon representation which is coloured from red (N-terminus) to blue (C-terminus). [2Fe2S] cluster and parts of the four anchoring cysteine residues (carbon and sulphur atoms) are depicted in ball-and- stick fashion. The figure is generated with VMD [63] based on structure of a single aa mutant PDB:1A70 [116]. The results from these studies are highly encouraging, and the logical next step is to move on to a more complex system with more biological function. The [2Fe2S] ferredoxins are ubiquitous in biological systems and the plant-type [2Fe2S] ferredoxin [117] looks like a promising candidate for single-molecule 55 4.2. Structure of a plant-type [2Fe2S] ferredoxin AFM studies (see Fig. 4.1). This type of ferredoxin adopts the so called β-grasp or UB-roll folding motif [99, 100, 117]. Some well known mechanically strong proteins (including: ubiquitin [77], GB1 [28, 90], and protein L [88]) also adopt this particular folding motif. Despite of the similarity in folding motif, the ferredoxin is functionally unre- lated to all of these mechanical strong proteins. With the [2Fe2S] cluster carried on a loop, ferredoxin serves as an electron carrier in diverse metabolic pathways [117]. In the case of higher plants, [2Fe2S] ferredoxin of type I works as an im- portant component in photosynthesis [117]. The protein is encoded by a nuclear gene and is produced in the cytoplasm as a larger precursor protein. The precursor is translocated into the chloroplast for proteolytic processing and cluster assembly to yield the mature ferredoxin [118–120]. The biogenesis of iron-sulphur protein is a topic under active research [121] (see Fig. 4.2 and the reference). Figure 4.2: Biogenesis of iron-sulphur proteins in vivo. Inorganic sulphur atoms in the iron-sulphur cluster are derived from desulphurase mediated con- version of cysteint into alanine. An iron-sulphur cluster is first assembled onto a scaffold protein, it is later transferred to the target apo-protein with the help of additional transfer proteins. The target apo-protein, together with correctly as- sembled and bound iron-sulphur cluster, becomes a mature iron-sulphur protein. Adapted by permission from Macmillan Publishers Ltd: Nature [121] c©(2009). 56 4.2. Structure of a plant-type [2Fe2S] ferredoxin Figure 4.3: UV-visible absorption spectra of spinach ferredoxin. Spectra are recorded at 0, 6, 12, and 24 hours of incubation at room temperature in 10 mM Tris-HCl buffer (pH 7.4). Inset graph shows ratio of absorbances at 465 nm and 276 nm (A465/A276) plotted against time. The ratio is essentially constant in the presence of 1 M NaCl. Contrast this to the decrease in this ratio in the absence of NaCl. Reprinted from [122], c©(1979), with permission from Elsevier. The iron-sulphur cofactor not only gives ferredoxin its function, but may also serve as a structural “staple” that help maintain the structure of the protein. Elec- tronic structure of the [2Fe2S] cluster dictates that electronic transition arising from ligand to metal charge transfer (sulphur to iron [123]) falls in the visible 57 4.3. Mechanical design of spinach ferredoxin region (∼300-600 nm) of the UV-visible spectrum. Thus, UV-visible absorption spectroscopy is a traditional ensemble level experimental technique commonly used in studies of iron-sulphur proteins. Absorption spectrum (see Fig. 4.3, 0 hour incubation) of a typical (and often studied) plant-type [2Fe2S] ferredoxin, spinach ferredoxin I (FdS for short), contains four (4) significant absorption max- ima at ∼280 nm, 330 nm, 420 nm, and 460 nm [117]. The maxima at ∼280 nm and ∼420 nm correspond to absorption by the aromatic aa of the protein scaffold and [2Fe2S] cluster, respectively. The ratio of absorbances at these wavelengths (A420/A280) is often used as a measure of the fraction of holo-protein (holo refers to the mature protein, i.e. protein with cofactor) [117]. For FdS, this ratio is usu- ally ∼0.5 and a FdS sample with A420/A280 = 0.5 is often taken to be essentially 100% pure holo-protein [110, 113, 124]. In a study published in 1979, Hasumi and colleagues report on the thermal stability of FdS under incubation at room temperature [122]. Degradation of holo- FdS is followed by UV-visible absorption spectroscopy (see Fig. 4.3) [122]. An increasingly pronounced bleaching of absorption maxima in the visible region has been observed as incubation time increases. The researchers have attributed this to the loss of [2Fe2S] cluster and generation of apo-FdS (apo refers to the pro- tein scaffold, i.e. protein without cofactor). Moreover, these authors and other researchers have noted that the generation of apo-FdS is almost always accom- panied with significant loss of protein structure, based on evidence from circular dichroism spectroscopy [122, 125]. 4.3 Mechanical design of spinach ferredoxin Structural features of FdS make it an interesting subject to study in mechanical denaturation. The β-grasp folding motif implies FdS may also be a mechanically strong protein like GB1 and ubiquitin. Moreover, the [2Fe2S] cluster may also exhibit significant mechanical stability like the [FeS4] centre in rubredoxin. Mature FdS has 97 aa in total. The [2Fe2S] cluster is covalently attached to 58 4.3. Mechanical design of spinach ferredoxin Figure 4.4: Coordiantion sphere of iron in the cluster. Covalently bound [2Fe2S] cluster of FdS depicted in ball-and-stick fashion with aa number of the sulphur atoms from cysteine residues labeled. The figure is generated with VMD [63] based on structure from PDB:1A70 [116]. the protein through cysteine residues at aa positions 39, 44, 47, and 77 (see Fig. 4.4). If one assumes both FdS and its [2Fe2S] cluster are mechanically strong, what will its mechanical unfolding signature be? Since the structure of wt FdS is unavailable on PDB, mechanical design of FdS will be analyzed based on the structure of a mutant (aa 92 is mutated from glutamic acid to a lysine) [116]. Contour length of the unfolded FdS polypeptide is ∼34.9 nm (0.36 nm/aa × 97 aa). Euclidean distance between the N-C termini Cα atoms is determined to be ∼3.4 nm using VMD [63] based on structure from PDB:1A70 [116]. Contour length increment for the complete mechanical unfolding is ∼31.5 nm. If one as- sumes the mechanical unfolding proceed in two steps: rupture of protein part and rupture of cluster part, then one can divide the contour length increment in two parts. Effective contour length increment from rupture of the cluster will be ∼13.2 nm, Since there are 39 aa from the first (aa number 39) to last (aa number 77) an- choring cysteine residues. Consequently, contour length increment from rupture of protein part will be ∼18.3 nm. 59 4.4. Rupture of protein structure and [2Fe2S] cluster under force 4.4 Rupture of protein structure and [2Fe2S] cluster under force Before we can carry out constant-velocity SMFS experiments to investigate the mechanical design of FdS, a polyprotein needs to be produced. Our goal in the protein engineering step is to produce a polyprotein with a known number of ma- ture FdS domains. In order to achieve this, one must keep in mind the general idea of biogenesis of iron-sulphur proteins (see Fig. 4.2). The biogenesis occurs in two general steps: production of FdS polypeptide scaffold as an apo-protein and enzyme-assisted [2Fe2S] cluster insertion onto the scaffold. Traditional polyprotein construction is carried out at gene level [62]. Gene en- coding FdS domain will have to be engineered into tandem repeats using standard molecular biology techniques. Naturally, the polypeptide synthesized from the polyprotein gene contains tandem repeats of aa sequences corresponding to tan- dem repeats of apo-FdS. Since the enzymatic machineries in a biological system is not designed to handle tandem repeats of apo-FdS, one anticipates the [2Fe2S] cluster insertion onto tandem repeats of apo FdS may be problematic. In order to avoid any unnecessary complications, we have chosen to produce bi-cysteine mutants of FdS and proceed through the chemical coupling route to- wards the production of a polyprotein. (see Ch. 3 and reference [51]). Addition of cysteine residues at the N-C termini seems to be the intuitive choice. However, our attempt to generate chemical cross-linked polyprotein from such a construct has failed based on evidence from AFM experiments. We never observe signifi- cant number of long polyproteins (e.g. more than two domains of FdS) in AFM experiments. The cause of this failure is obscure at this stage. We are tempted to think this may be due to possible structural flexibility in the small C-terminal α-helix, since Binda and colleagues have only crystallized a point mutant which bears a mutation on the α-helix [116]. Instead of addition of a cysteine residue after the C-terminus, we have decided to mutate aa number 93 from glutamic acid to cysteine. Literature reports suggest 60 4.4. Rupture of protein structure and [2Fe2S] cluster under force the three glutamic acid residues on the small α-helix may be involved in protein- protein interactions between FdS and its redox partners [126]. Following the work of Binda and colleagues [116], we have chosen to move the C-terminal cysteine to a residue on the small C-terminal α-helix. Glutamic acid at aa number 93 was chosen since it is not believe to be involved in any interactions stabilizing the protein [116, 126]. It has been reported the homologous glutamic acid at aa number 95 in [2Fe2S] ferredoxin from a cynobacteria strain, vegetative Anabaena 7120, is solvent exposed [127]. The bi-cysteine mutant is referred to as cF93c for brevity. Figure 4.5: Mechanical design of FdS. Typical force versus extension traces from constant-velocity single-molecule AFM experiments are presented in panel (A). Dotted lines are fits using the interpolation force versus extension formula of the WLC model (see equation 1.1 and [101]). Typical persistence length p used is ∼0.4 nm. Normalized frequency histograms of rupture forces are presented in panel (B). Gaussian fits plotted as solid curves are shown to guide the eye. Protein engineering and chemical coupling reaction of the cF93c mutant is car- 61 4.4. Rupture of protein structure and [2Fe2S] cluster under force ried out as described in subsections 4.1.1 and 4.1.2, respectively. Assuming the two anchoring cysteine residues have minimal effects on FdS other than serving as bridges between two domains, we use the names FdS and cF93c interchange- ably. Constant-velocity single-molecule AFM experiments reveal FdS is indeed a mechanically strong protein. Its mechanical unfolding proceeds in two steps: an ∼18 nm step corresponding to rupture of protein part of FdS and an ∼13 nm step corresponding to rupture of cluster part of FdS (e.g. see top graph in Fig. 4.5 A). For a single domain of mature holo-FdS, the protein rupture step always precedes the cluster rupture step. Folding topology dictates the mechanical clamp in the protein part of FdS must be overpowered before the [2Fe2S] cluster is subjected to mechanical stress. This idea is very similar to design concepts used in previous loop [109] and domain [128] insertion studies. One immediate consequence of such a design is the so-called reversal of mechanical unfolding hierarchy [128]. The normal mechanical unfolding hierarchy in a tandem modular polyprotein is unfolding proceeds from the mechanically weakest to the strongest. Mechanically labile proteins (e.g. random coils and proteins with mostly α-helix) will unfolding before mechanically strong proteins (e.g. proteins with mostly β-sheets). Folding topology dictates that protein rupture must precede cluster rupture, regardless of their relative mechanical strength. This is exactly what we observe in our single-molecule AFM experiments. In top graph in Fig. 4.5 A, each protein rupture step (∼18 nm) is followed by a cluster rupture step (∼13 nm). By visual estimates from the graph, cluster rupture happens at a low force than protein rup- ture in general. The ratio of protein rupture step to cluster rupture step is exactly 1:1 in the current example, this is not the case in the example in bottom graph in Fig. 4.5 A. In general, one can not expect this ratio to be exactly 1:1 since it is affected by a number of factors including: detachment time of molecule, rupture events masked by non-specific interactions, and ratio of holo-FdS to apo-FdS. One can not have direct control over the first two factors, but their influences on the apparent ratio of rupture steps tend to be averaged out by gathering good statistics. Ratio of holo-FdS to apo-FdS is determined by our sample preparation procedure. 62 4.4. Rupture of protein structure and [2Fe2S] cluster under force As discussed in section 4.2, UV-Visible absorption characteristics of FdS is often used to estimate the holo:apo ratio. An A420/A280 ratio of ∼0.5 is often taken to be 100% holo-FdS. Our sample preparation procedure can consistently yield FdS with A420/A280 ≈ 0.4 (see Fig. 4.6), and we estimate the holo:apo ratio to be ∼4:1 (i.e. 80% holo-FdS). Please read appendix A for more spectra (see Fig. A.1, Fig. A.2, and Fig. A.3). Figure 4.6: UV-visible absorption spectrum of cF93c. We have carried out multiple constant-velocity single-molecule AFM exper- iments to collect statistics on protein and cluster rupture steps. Normalized fre- quency histograms of unfolding forces are displayed in Fig. 4.5 B. The rupture force of the protein rupture step is measured to be 179± 43 pN (mean ± SD) from 317 events, whereas that of the cluster rupture step is measured to be 153± 48 pN from 359 events. There are ∼13% more cluster rupture events observed. Assum- ing the sampling is unbiased, this suggests a small fraction of the total population exists in a partially-denatured holo-form. In this form, the protein rupture step is absent and the cluster rupture step is still present. Until this stage, the apo-FdS 63 4.4. Rupture of protein structure and [2Fe2S] cluster under force (∼20%) fraction of the population has not been addressed. It is possible for the apo-FdS to exists in two forms: a partially-denatured form and a pseudo-native form. The pseudo-native form should be able to resist mechanical stress while the partially-denatured form should not. If the pseudo-native form exists, then the contour length increment for it will be ∼31 nm. Figure 4.7: Iron-sulphur cluster and mechanical design of FdS. Typical force versus extension traces from constant-velocity single-molecule AFM experiments are shown: no EDTA added (panel A) and 50 mM EDTA, ∼12 hours incubation at room temperature (panel B). Since our sample preparation procedure produces mostly holo-FdS (∼80%), a method to generate a significantly fraction of apo-FdS from holo-FdS is required. Incubation of holo-FdS in the presence of a metal chelator like EDTA at room temperature has been suggested as a mild method for the generation of apo-FdS from holo-FdS [129]. Since the [2Fe2S] cluster is covalently bound to the protein scaffold, one expects the insertion and removal kinetics to be slow (on the time scale of ∼few hours [129–131]). We would like to carry out constant-velocity single-molecule AFM experiments in the present of a metal chelator like EDTA to distinguish the two possible apo-FdS species. However, one anticipates a problem 64 4.4. Rupture of protein structure and [2Fe2S] cluster under force if the partially-denatured form is the dominate apo-FdS species. Because there will not be a mechanical signature if both the protein and cluster rupture steps are absent. To overcome this obstacle, a mechanical fingerprint domain, GL5 is introduced into the protein design to give a hetero-dimeric design (cGl5-F93c). Globular protein domain GL5 is a loop insertion mutant of GB1 and it has been characterized extensively [109, 128]. GL5 unfolds at 134 ± 36 pN under 400 nm/s pulling velocity and has a contour length increment of ∼ 20 nm [109, 128]. Assuming EDTA does not have any significant effects on the mechanical strength of GL5, it can serve as a mechanical fingerprint and internal force calliper. Protein engineering and chemical coupling reaction of the cGL5-F93c mutant is carried out as described in subsections 4.1.1 and 4.1.2, respectively. Typical force versus extension traces of GL5-FdS (this will be used interchangeably with cGL5-F93c), before and after incubation with EDTA are shown in Fig. 4.7. Vi- sual inspection of the data reveals a slight decrease in the relative number of FdS rupture events after ∼12 hours of incubation in 50 mM EDTA. More importantly, rupture events with ∆Lc ≈ 31 nm almost never appears. In collaboration of a col- league in the lab, multiple contant-velocity single-molecule AFM experiments are currently underway to collect good statistics on the GL5-FdS hetero-dimer. When enough data is collect to build unfolding force histograms, more information about apo-FdS and the effect of EDTA can be learned. Mechanical denaturation experiments using single-molecule AFM have shed light on the mechanical design of a plant-type [2Fe2S] ferredoxin. It is shown that mechanical unfolding of native holo-FdS proceeds in two steps with pro- tein rupture followed by cluster rupture. Removal of the iron-sulphur cluster has profound effects on the mechanical stability of FdS since a mechanically strong pseudo-native apo-FdS is never observed. Our current study nicely compliments previous studies on the role of iron-sulphur cluster in the stability of FdS [125]. Our preliminary results demonstrate the use of single-molecule AFM in the study of iron-sulphur proteins and open avenues to new experiments. 65 Chapter 5 Conclusions and Future Prospects In chapter 3, a small globular protein GB1 has been shown to exhibit pronounced anisotropic response to directional mechanical stress. The same protein can be both mechanically strong and weak depending on the direction of the applied force. When the direction of the applied force is aligned with the intrinsic long axis of the β-sheet, GB1 is mechanically strong. This has been attributed to the concerted disruption of native contacts. When the direction of the applied force is aligned perpendicular to the the intrinsic long axis of the β-sheet, GB1 is mechani- cally weak. This has been attributed to the sequential disruption of native contacts. Knowledge of mechanical anisotropy may aid the design of protein based novel elastic materials. In chapter 4, the mechanical design of a metalloprotein has been characterized. The [2Fe2S] ferredoxin from spinach has similar folding topology to GB1. It is shown that both the protein and iron-sulphur cluster are mechanically strong and mechanical unfolding of the ferredoxin proceeds in two steps. Folding topology dictates protein rupture always precedes cluster rupture in the case of the native holo-ferredoxin. Removal of the cluster by a metal chelator also abolishes the mechanical strength of the protein part. Function and structure are sagaciously intertwined in this simple iron-sulphur protein. In this thesis, SMFS has been carried out using the atomic force microscope to investigate the mechanical design of protein molecules. Mechanical denaturation is achieved via an applied mechanical force. SMFS experiments work in concert with protein engineering and SMD simulations to provided single-molecule infor- mation. Technical advances in experiments [132] and simulations [133] will open doors to more sophisticated and interesting experiments in the future. 66 Appendix A Additional UV-Visible Spectra Additional UV-visible spectra are included for reference. Figure A.1: UV-visible absorption spectrum of concentrated cF93c. The A420/A280 ratio is reduced to ∼0.37 from ∼0.40 in Fig. 4.6. 67 Appendix A. Additional UV-Visible Spectra Figure A.2: UV-visible absorption spectra. Top: UV-visible spectrum of the same sample as in Fig. A.1 after ∼1 week storage at 4◦C. The A420/A280 ratio is further reduced to ∼0.30. Bottom: UV-visible spectrum of cGL5-F93c. The A420/A280 ratio is ∼0.20 but the decrease in this ratio is partially due to the in- creased molar extinction coefficient at ∼280 nm. 68 Appendix A. Additional UV-Visible Spectra Figure A.3: UV-visible absorption spectra in the presence/absence of EDTA. Top: incubated at room temperature for ∼2 days without EDTA. Bottom: incu- bated at room temperature for ∼2 days with 5 mM EDTA. The A420/A280 ratio is reduced by more than 90%. 69 Bibliography [1] Binnig, G., Quate, C. F. & Gerber, C. Atomic force microscope. Phys. Rev. Lett. 56, 930–933 (1986). [2] Alessandrini, A. & Facci, P. AFM: a versatile tool in biophysics. Meas. Sci. Technol. 16, R65–R92 (2005). [3] Marszalek, P. E. & Dufrêne, Y. F. Stretching single polysaccharides and proteins using atomic force microscopy. Chem. Soc. Rev. 41, 3523–3534 (2012). [4] Müller, D. J., Helenius, J., Alsteens, D. & Dufrêne, Y. F. Force probing surfaces of living cells to molecular resolution. Nat. Chem. Biol. 5, 383– 390 (2009). [5] Dufrêne, Y. F. et al. Five challenges to bringing single-molecule force spectroscopy into living cells. Nat. Methods 8, 123–127 (2011). [6] Forman, J. R. & Clarke, J. Mechanical unfolding of proteins: insights into biology, structure and folding. Curr. Opin. Struct. Biol. 17, 58–66 (2007). [7] Žoldák, G. & Rief, M. Force as a single molecule probe of multidi- mensional protein energy landscapes. Curr. Opin. Struct. Biol. 23, 48–57 (2012). [8] Crampton, N. & Brockwell, D. J. Unravelling the design principles for single protein mechanical strength. Curr. Opin. Struct. Biol. 20, 508–517 (2010). 70 Bibliography [9] Li, H. B. & Cao, Y. Protein mechanics: from single molecules to functional biomaterials. Acc. Chem. Res. 43, 1331–1341 (2010). [10] Alberts, B. et al. Molecular biology of the cell (Garland Science, New York, 2002), 4th edn. [11] Berg, J., Tymoczko, J. L. & Stryer, L. Biochemistry (W. H. Freeman, New York, 2002), 5th edn. [12] Chang, R. Physical chemistry for the biosciences (University Science Books, 2005). [13] Bartlett, A. I. & Radford, S. E. An expanding arsenal of experimental methods yields an explosion of insights into protein folding mechanisms. Nat. Struct. Mol. Biol. 16, 582–588 (2009). [14] Onuchic, J. N. & Wolynes, P. G. Theory of protein folding. Curr. Opin. Struct. Biol. 14, 70–75 (2004). [15] Dobson, C. M. Protein folding and misfolding. Nature 426, 884–890 (2003). [16] Zwanzig, R., Szabo, A. & B., B. Levinthal’s paradox. Proc. Natl. Acad. Sci. USA 89, 20–22 (1992). [17] Vogel, V. Mechanotransduction involving multimodular proteins: convert- ing force into biochemical signals. Annu. Rev. Biophys. Biomol. Struct. 35, 459–488 (2006). [18] Bustamante, C., Chemla, Y. R., Forde, N. R. & Izhaky, D. Mechanical processes in biochemistry. Annu. Rev. Biochem. 73, 705–748 (2004). [19] Puchner, E. M. & Gaub, H. E. Single-molecule mechanoenzymatics. Annu. Rev. Biophys. 41, 497–518 (2012). 71 Bibliography [20] Neuman, K. C., Lionnet, T. & Allemand, J. F. Single-molecule microma- nipulation techniques. Annu. Rev. Mater. Res. 37, 33–67 (2007). [21] Neuman, K. C. & Nagy, A. Single-molecule force spectroscopy: optical tweezers, magnetic tweezers and atomic force microscopy. Nat. Methods 5, 491–505 (2008). [22] Cappella, B. & Dietler, G. Force-distance curves by atomic force mi- croscopy. Surf. Sci. Rep. 34, 1–104 (1999). [23] Butt, H. J., Cappella, B. & Kappl, M. Force measurements with the atomic force microscope: technique, interpretation and applications. Surf. Sci. Rep. 59, 1–152 (2005). [24] Carrion-Vazquez, M. et al. Mechanical design of proteins: studied by single-molecule force spectroscopy and protein engineering. Prog. Bio- phys. Mol. Biol. 74, 63–91 (2000). [25] Rief, M., Gautel, M., Oesterhelt, F., Fernandez, J. M. & Gaub, H. E. Re- versible unfolding of individual titin immunoglobulin domains by AFM. Science 276, 1109–1112 (1997). [26] Rief, M. & Grubmuller, H. Force spectroscopy of single biomolecules. ChemPhysChem 3, 255–261 (2002). [27] Li, H. B., Oberhauser, A. F., Fowler, S. B., Clarke, J. & Fernandez, J. M. Atomic force microscopy reveals the mechanical design of a modular pro- tein. Proc. Natl. Acad. Sci. USA 97, 6527–6531 (2000). [28] Cao, Y. & Li, H. B. Polyprotein of GB1 is an ideal artificial elastomeric protein. Nat. Mater. 6, 109–114 (2007). [29] Izrailev, S., Stepaniants, S., Balsera, M., Oono, Y. & Schulten, K. Molecu- lar dynamics study of unbinding of the avidin-biotin complex. Biophys. J. 72, 1568–1581 (1997). 72 Bibliography [30] Flory, P. J. Statistical mechanics of chain molecules (Interscience Publish- ers, New York, 1969). [31] van der Maarel, J. R. C. Introduction to Biopolymer Physics (World Scien- tific Publishing Company, Singapore, 2008). [32] Rounsevell, R., Forman, J. R. & Clarke, J. Atomic force microscopy: me- chanical unfolding of proteins. Methods 34, 100–111 (2004). [33] Zwanzig, R. Two-state models of protein folding kinetics. Proc. Natl. Acad. Sci. USA 94, 148–150 (1997). [34] Bell, G. I. Models for specific adhesion of cells to cells. Science 200, 618–627 (1978). [35] Evans, E. & Ritchie, K. Dynamic strength of molecular adhesion bonds. Biophys. J. 72, 1541–1555 (1997). [36] Hummer, G. & Szabo, A. Kinetics from nonequilibrium single-molecule pulling experiments. Biophys. J. 85, 5–15 (2003). [37] Dudko, O. K., Hummer, G. & Szabo, A. Intrinsic rates and activation free energies from single-molecule pulling experiments. Phys. Rev. Lett. 96, 108101 (2006). [38] Dudko, O. K., Hummer, G. & Szabo, A. Theory, analysis, and interpreta- tion of single-molecule force spectroscopy experiments. Proc. Natl. Acad. Sci. USA. 105, 15755–15760 (2008). [39] Kellermayer, M. S. Z., Smith, S. B., Granzier, H. L. & Bustamante, C. Folding-unfolding transitions in single titin molecules characterized with laser tweezers. Science 276, 1112–1116 (1997). [40] Tskhovrebova, L., Trinick, J., Sleep, J. A. & Simmons, R. M. Elasticity and unfolding of single molecules of the giant muscle protein titin. Nature 387, 308–312 (1997). 73 Bibliography [41] Rief, M., Pascual, J., Saraste, M. & Gaub, H. E. Single molecule force spectroscopy of spectrin repeats: low unfolding forces in helix bundles. J. Mol. Biol. 286, 553–561 (1999). [42] Lu, H., Isralewitz, B., Krammer, A., Vogel, V. & Schulten, K. Unfolding of titin immunoglobulin domains by steered molecular dynamics simulation. Biophys. J. 75, 662–671 (1998). [43] Paci, E. & Karplus, M. Unfolding proteins by external forces and temper- ature: the importance of topology and energetics. Proc. Natl. Acad. Sci. USA 97, 6521–6526 (2000). [44] Klimov, D. K. & Thirumalai, D. Native topology determines force-induced unfolding pathways in globular proteins. Proc. Natl. Acad. Sci. USA 97, 7254–7259 (2000). [45] Dietz, H. et al. Cysteine engineering of polyproteins for single-molecule force spectroscopy. Nat. Protoc. 1, 80–84 (2006). [46] Dietz, H., Berkemeier, F., Bertz, M. & Rief, M. Anisotropic deforma- tion response of single protein molecules. Proc. Natl. Acad. Sci. USA 103, 12724–12728 (2006). [47] Cao, Y., Yoo, T. & Li, H. B. Single molecule force spectroscopy reveals engineered metal chelation is a general approach to enhance mechanical stability of proteins. Proc. Natl. Acad. Sci. USA 105, 11152–11157 (2008). [48] Zhuang, S., Peng, Q., Cao, Y. & Li, H. B. Modulating the mechanical stability of extracellular matrix protein tenascin-C in a controlled and re- versible fashion. J. Mol. Biol. 390, 820–829 (2009). [49] Cao, Y., Er, K. S., Parhar, R. & Li, H. B. A force-spectroscopy-based single-molecule metal-binding assay. ChemPhysChem 10, 1450–1454 (2009). 74 Bibliography [50] Cao, Y. & Li, H. B. Dynamics of protein folding and cofactor binding monitored by single-molecule force spectroscopy. Biophys. J. 101, 2009– 2017 (2011). [51] Zheng, P., Cao, Y. & Li, H. B. Facile method of constructing polyproteins for single-molecule force spectroscopy studies. Langmuir 27, 5713–5718 (2011). [52] Nölting, B. Methods in modern biophysics (Springer, New York, 2009), 3rd edn. [53] Meyer, G. & Amer, N. M. Novel optical approach to atomic force mi- croscopy. Appl. Phys. Lett. 53, 1045–1047 (1988). [54] Meyer, E., Hug, H. J. & Bennewitz, R. Scanning probe microscopy: the lab on a tip (Springer, New York, 2003). [55] Morris, V. J., Gunning, A. P. & Kirby, A. R. Atomic force microscopy for biologists (World Scientific Publishing Company, Singapore, 1999). [56] Wang, C. & Bai, C. Single molecule chemistry and physics: an introduction (Springer, New York, 2006). [57] te Riet, J. et al. Inter-laboratory round robin on cantilever calibration for AFM force spectroscopy. Ultramicroscopy 111, 1659–1669 (2011). [58] Carrion-Vazquez, M. et al. Advanced Techniques in Biophysics, vol. 10 of Springer Series in Biophysics, chap. Protein nanomechanics - as studied by AFM single-molecule force spectroscopy (Springer, 2006). [59] Hutter, J. L. & Bechhoefer, J. Calibration of atomic-force microscope tips. Rev. Sci. Instrum. 64, 1868–1873 (1993). [60] Florin, E. L. et al. Sensing specific molecular interactions with the atomic force microscope. Biosens. and Bioelectron. 10, 895–901 (1995). 75 Bibliography [61] Fisher, T. E., Oberhauser, A. F., Carrion-Vazquez, M., Marszalek, P. E. & Fernandez, J. M. The study of protein mechanics with the atomic force microscope. Trends Biochem. Sci. 24, 379–384 (1999). [62] Hoffmann, T. & Dougan, L. Single molecule force spectroscopy using polyproteins. Chem. Soc. Rev. 41, 4781–4796 (2012). [63] Humphrey, W., Dalke, A. & Schulten, K. VMD: visual molecular dynam- ics. J. Mol. Graphics 14, 33–38 (1996). [64] Fisher, T. E. et al. Single molecule force spectroscopy of modular proteins in the nervous system. Neuron 27, 435–446 (2000). [65] Evans, E. & Ritchie, K. Strength of a weak bond connecting flexible poly- mer chains. Biophys. J. 76, 2439–2447 (1999). [66] Voter, A. F. Radiation Effects in Solids, vol. 235 of NATO Science Se- ries, chap. Introduction to the kinetic Monte Carlo method, 1–23 (Springer, 2007). [67] Gillespie, D. T. Stochastic simulation of chemical kinetics. Annu. Rev. Phys. Chem. 58, 35–55 (2007). [68] Rief, M., Fernandez, J. M. & Gaub, H. E. Elastically coupled two-level systems as a model for biopolymer extensibility. Phys. Rev. Lett. 81, 4764– 4767 (1998). [69] Oberhauser, A. F., Marszalek, P. E., Erickson, H. P. & Fernandez, J. M. The molecular elasticity of the extracellular matrix protein tenascin. Nature 393, 181–185 (1998). [70] Carrion-Vazquez, M. et al. Mechanical and chemical unfolding of a single protein: a comparison. Proc. Natl. Acad. Sci. USA 96, 3694–3699 (1999). 76 Bibliography [71] Press, W., Teukolsky, S., Vetterling, W. & Flannery, B. Numerical recipes in Fortran: the art of scientific computing (Cambridge university press, Cambridge, UK, 1992), 2nd edn. [72] Matsumoto, M. & Nishimura, T. Mersenne twister: a 623-dimensionally equidistributed uniform pseudo-random number generator. ACM Trans. Model. Comput. Simul. 8, 3–30 (1998). [73] Brooks, B. R. et al. CHARMM: a program for macromolecular energy, minimization, and dynamics calculations. J. Comput. Chem. 4, 187–217 (1983). [74] Brooks, B. R. et al. CHARMM: the biomolecular simulation program. J. Comput. Chem. 30, 1545–1614 (2009). [75] Borrell, B. Chemistry: power play. Nature 451, 240–243 (2008). [76] Sotomayor, M. & Schulten, K. Single-molecule experiments in vitro and in silico. Science 316, 1144–1148 (2007). [77] Carrion-Vazquez, M. et al. The mechanical stability of ubiquitin is linkage dependent. Nat. Struct. Biol. 10, 738–743 (2003). [78] Brockwell, D. J. et al. Pulling geometry defines the mechanical resistance of a beta-sheet protein. Nat. Struct. Biol. 10, 731–737 (2003). [79] Li, P. C., Huang, L. & Makarov, D. E. Mechanical unfolding of segment- swapped protein G dimer: results from replica exchange molecular dynam- ics simulations. J. Phys. Chem. B 110, 14469–14474 (2006). [80] Nome, R. A., Zhao, J. M., Hoff, W. D. & Scherer, N. F. Axis-dependent anisotropy in protein unfolding from integrated nonequilibrium single- molecule experiments, analysis, and simulation. Proc. Natl. Acad. Sci. USA 104, 20799–20804 (2007). 77 Bibliography [81] Glyakina, A. V., Balabaev, N. K. & Galzitskaya, O. V. Multiple unfolding intermediates obtained by molecular dynamic simulations under stretching for immunoglobulin-binding domain of protein G. Open Biochem. J. 3, 66–77 (2009). [82] Glyakina, A. V., Balabaev, N. K. & Galzitskaya, O. V. Mechanical unfold- ing of proteins L and G with constant force: similarities and differences. J. Chem. Phys. 131, 045102 (2009). [83] Lee, W. et al. Mechanical anisotropy of ankyrin repeats. Biophys. J. 102, 1118–1126 (2012). [84] Nowak, W. & Marszalek, P. E. Computational chemistry: reviews of current trends, vol. 9, chap. Molecular Dynamics Simulations of Single Molecule Atomic Force Microscope Experiments, 47–83 (World Scientific Publishing Company, 2005). [85] Genchev, G. Z. et al. Mechanical signaling on the single protein level stud- ied using steered molecular dynamics. Cell Biochem. Biophys. 55, 141–152 (2009). [86] Li, Y. D., Lamour, G., Gsponer, J., Zheng, P. & Li, H. B. The molecular mechanism underlying mechanical anisotropy of the protein GB1. Biophys. J. 103, 2361–2368 (2012). [87] Ng, S. P. et al. Designing an extracellular matrix protein with enhanced mechanical stability. Proc. Natl. Acad. Sci. USA 104, 9633–9637 (2007). [88] Sadler, D. P. et al. Identification of a mechanical rheostat in the hydropho- bic core of protein L. J. Mol. Biol. 393, 237–248 (2009). [89] Yang, G. L. et al. Solid-state synthesis and mechanical unfolding of poly- mers of T4 lysozyme. Proc. Natl. Acad. Sci. USA 97, 139–144 (2000). 78 Bibliography [90] Cao, Y., Lam, C., Wang, M. J. & Li, H. B. Nonmechanical protein can have significant mechanical stability. Angew. Chem. Int. Ed. 45, 642–645 (2006). [91] Graham, T. G. W. & Best, R. B. Force-induced change in protein unfolding mechanism: Discrete or continuous switch? J. Phys. Chem. B 115, 1546– 1561 (2011). [92] Fernandez, J. M. & Li, H. B. Force-clamp spectroscopy monitors the fold- ing trajectory of a single protein. Science 303, 1674–1678 (2004). [93] MacKerell, A. D. et al. All-atom empirical potential for molecular mod- eling and dynamics studies of proteins. J. Phys. Chem. B 102, 3586–3616 (1998). [94] Mackerell, A. D., Feig, M. & Brooks, C. L. Extending the treatment of backbone energetics in protein force fields: limitations of gas-phase quantum mechanics in reproducing protein conformational distributions in molecular dynamics simulations. J. Comput. Chem. 25, 1400–1415 (2004). [95] Haberthur, U. & Caflisch, A. FACTS: fast analytical continuum treatment of solvation. J. Comput. Chem. 29, 701–715 (2008). [96] Ryckaert, J. P., Ciccotti, G. & Berendsen, H. J. C. Numerical-integration of Cartesian equations of motion of a system with constraints: molecular- dynamics of n-alkanes. J. Comput. Phys. 23, 327–341 (1977). [97] Gronenborn, A. M. et al. A novel, highly stable fold of the immunoglobulin binding domain of streptococcal protein G. Science 253, 657–661 (1991). [98] Gallagher, T., Alexander, P., Bryan, P. & Gilliland, G. L. Two crystal- structures of the B1 immunoglobulin-binding domain of streptococcal pro- tein G and comparison with NMR. Biochemistry 33, 4721–4729 (1994). 79 Bibliography [99] Overington, J. P. Comparison of three-dimensional structures of homolo- gous proteins. Curr. Opin. Struct. Biol. 2, 394–401 (1992). [100] Orengo, C. Classification of protein folds. Curr. Opin. Struct. Biol. 4, 429–440 (1994). [101] Marko, J. F. & Siggia, E. D. Stretching DNA. Macromolecules 28, 8759– 8770 (1995). [102] Cao, Y. & Li, H. B. Engineered elastomeric proteins with dual elasticity can be controlled by a molecular regulator. Nat. Nanotechnol. 3, 512–516 (2008). [103] West, D. K., Brockwell, D. J., Olmsted, P. D., Radford, S. E. & Paci, E. Mechanical resistance of proteins explained using simple molecular mod- els. Biophys. J. 90, 287–297 (2006). [104] Sharma, D. et al. Stabilization provided by neighboring strands is critical for the mechanical stability of proteins. Biophys. J. 95, 3935–3942 (2008). [105] Sharma, D. et al. Single-molecule force spectroscopy reveals a mechani- cally stable protein fold and the rational tuning of its mechanical stability. Proc. Natl. Acad. Sci. USA 104, 9278–9283 (2007). [106] Shen, T., Cao, Y., Zhuang, S. & Li, H. B. Engineered bi-histidine metal chelation sites map the structure of the mechanical unfolding transition state of an elastomeric protein domain GB1. Biophys. J. 103, 807–816 (2012). [107] Wittung-Stafshede, P. Role of cofactors in protein folding. Acc. Chem. Res. 35, 201–208 (2002). [108] Wilson, C. J., Apiyo, D. & Wittung-Stafshede, P. Role of cofactors in metalloprotein folding. Q. Rev. Biophys. 37, 285–314 (2004). 80 Bibliography [109] Li, H. B., Wang, H.-C., Cao, Y., Sharma, D. & Wang, M. Configurational entropy modulates the mechanical stability of protein GB1. J. Mol. Biol. 379, 871–880 (2008). [110] Petering, D. & Palmer, G. Properties of spinach ferredoxin in anaerobic urea solution: a comparison with the native protein. Arch. Biochem. Bio- phys. 141, 456–464 (1970). [111] Sakihama, N. et al. Toyopearl HW-65C: ammonium-sulfate as a new col- umn chromatographic adsorbent for enzyme-purification. J. Biochem. 93, 129–134 (1983). [112] Sakihama, N., Shin, M. & Toda, H. Separation and identification of two native forms of spinach ferredoxin by hydrophobic chromatography. J. Biochem. 100, 43–47 (1986). [113] Piubelli, L., Aliverti, A., Bellintani, F. & Zanetti, G. Spinach ferredoxin I: overproduction in Escherichia coli and purification. Protein Expr. Purif. 6, 298–304 (1995). [114] Zheng, P. & Li, H. B. Highly covalent ferric-thiolate bonds exhibit surpris- ingly low mechanical stability. J. Am. Chem. Soc. 133, 6791–6798 (2011). [115] Zheng, P., Takayama, S. J., Mauk, A. G. & Li, H. B. Hydrogen bond strength modulates the mechanical strength of ferric-thiolate bonds in rubredoxin. J. Am. Chem. Soc. 134, 4124–4131 (2012). [116] Binda, C., Coda, A., Aliverti, A., Zanetti, G. & Mattevi, A. Structure of the mutant E92K of [2Fe-2S] ferredoxin I from Spinacia oleracea at 1.7 Å resolution. Acta Crystallogr., Sect. D: Biol. Crystallogr. 54, 1353–1358 (1998). [117] Fukuyama, K. Structure and function of plant-type ferredoxins. Photo- synth. Res. 81, 289–301 (2004). 81 Bibliography [118] Takahashi, Y., Mitsui, A., Hase, T. & Matsubara, H. Formation of the iron sulfur cluster of ferredoxin in isolated chloroplasts. Proc. Natl. Acad. Sci. USA. 83, 2434–2437 (1986). [119] Takahashi, Y., Mitsui, A. & Matsubara, H. Formation of the Fe-S cluster of ferredoxin in lysed spinach chloroplasts. Plant Physiol. 95, 97–103 (1991). [120] Takahashi, Y., Mitsui, A., Fujita, Y. & Matsubara, H. Roles of ATP and NADPH in formation of the Fe-S cluster of spinach ferredoxin. Plant Phys- iol. 95, 104–110 (1991). [121] Lill, R. Function and biogenesis of iron-sulphur proteins. Nature 460, 831–838 (2009). [122] Hasumi, H., Nakamura, S., Koga, K. & Yoshizumi, H. Effects of neutral salts on thermal-stability of spinach ferredoxin. Biochem. Biophys. Res. Commun. 87, 1095–1101 (1979). [123] Lippard, S. J. & Berg, J. M. Principles of bioinorganic chemistry (Univer- sity Science Books, Mill Valley, CA, 1994). [124] Hasumi, H. & Ohmori, D. Studies on the molecular-structure of spinach ferredoxin. II. effects of urea and sodium-chloride. Biochim. Biophys. Acta 996, 173–180 (1989). [125] Pagani, S., Vecchio, G., Iametti, S., Bianchi, R. & Bonomi, F. On the role of the 2Fe-2S cluster in the formation of the structure of spinach ferredoxin. Biochim. Biophys. Acta 870, 538–544 (1986). [126] Aliverti, A., Livraghi, A., Piubelli, L. & Zanetti, G. On the role of the acidic cluster Glu 92-94 of spinach ferredoxin I. Biochim. Biophys. Acta 1342, 45–50 (1997). 82 Bibliography [127] Hurley, J. K. et al. Mutations of surface residues in Anabaena vegetative and heterocyst ferredoxin that affect thermodynamic stability as determined by guanidine-hydrochloride denaturation. Protein Sci. 4, 58–64 (1995). [128] Peng, Q. & Li, H. B. Domain insertion effectively regulates the mechanical unfolding hierarchy of elastomeric proteins: toward engineering multifunc- tional elastomeric proteins. J. Am. Chem. Soc. 131, 14050–14056 (2009). [129] Iametti, S. et al. Reversible, non-denaturing metal substitution in bovine adrenodoxin and spinach ferredoxin and the different reactivities of 2Fe-2S cluster-containing proteins. Eur. J. Biochem. 239, 818–826 (1996). [130] Crispin, D. J., Street, G. & Varey, J. E. Kinetics of the decomposition of [2Fe-2S] ferredoxin from spinach: implications for iron bioavailability and nutritional status. Food Chem. 72, 355–362 (2001). [131] Wu, S. P., Wu, G., Surerus, K. K. & Cowan, J. A. Iron-sulfur cluster biosyn- thesis. kinetic analysis of [2Fe-2S] cluster transfer from holo ISU to apo Fd: role of redox chemistry and a conserved aspartate. Biochemistry 41, 8876–8885 (2002). [132] Leitner, M. et al. Increased imaging speed and force sensitivity for bio- applications with small cantilevers using a conventional AFM setup. Mi- cron 43, 1399–1407 (2012). [133] Lee, E. H., Hsin, J., Sotomayor, M., Comellas, G. & Schulten, K. Discovery through the computational microscope. Structure 17, 1295–1306 (2009). 83

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0073778/manifest

Comment

Related Items