Analysis of Protein Denaturation Through Dynamic Light Scattering An Honors Thesis submitted in partial fulfillment of the requirement for Honors Studies in Physics By Jeffrey Andrew Sparks Spring 2003 Department of Physics J. William Fulbright College of Arts and Sciences The University of Arkansas Acknowledgements This project could not have happened without the tireless effort, support, guidance, instruction, prompting, reality doses, and collaboration of Dr. Lin Oliver of the University of Arkansas Physics Department. The honors theses and research of Nadeem Akbar, who designed the internal heater I used in my research, and April Fortner, who gave instrumental guidance on the production of protein, were invaluable as well. I also incorporated some of their figures in the experimental method portion and their strategy in deriving the theory of Dynamic Light Scattering. Dr. Wes Stites and Lee Manuel were also very helpful in offering advice and guidance during the long hours this physicist spent making protein. Finally, I would like to thank my family, friends, and international diversions who were always willing to keep my mind off this project. 2 Table of Contents TITLE PAGE...............................................................................1 ACKNOWLEDGEMENTS.........................................................2 TABLE OF CONTENTS ............................................................3 LIST OF FIGURES .....................................................................4 INTRODUCTION .......................................................................5 Proteins and Cell Biology: From DNA to Protein ..........................5 Protein Structure ..............................................................................8 The Problem With Folding ............................................................12 Protein Denaturation ......................................................................13 Methods of Protein Analyzation ....................................................15 Thermal Denaturation ....................................................................20 Goals ..............................................................................................22 THEORY OF DYNAMIC LIGHT SCATTERING ..................23 PROPERTIES OF STAPHYLOCOCCAL NUCLEASE .........30 EXPERIMENTAL METHOD ..................................................31 Production of Staphylococcal nuclease .........................................31 Dynamic Light Scattering Procedure .............................................32 RESULTS AND DISCUSSION................................................42 CONCLUSIONS .......................................................................52 REFERENCES ..........................................................................54 3 List of Figures Figure 1 - Generalized structure of an amino acid .........................................................7 Figure 2 - A typical peptide bond ..................................................................................8 Figure 3 - The structure of a sulfhydryl bond between cysteine groups ........................9 Figure 4 - A typical secondary structure of a protein, the ribbon α-helix....................10 Figure 5 - An example of the β-sheet, part of the secondary structure of a protein ....10 Figure 6 - The tertiary structure of ribulose bisphosphate carboxylase (RubisCo) .....11 Figure 7 - The structure of hemoglobin, a good example of quaternary protein structure............................................................................................................12 Figure 8 - Diagram of DLS incident and scattered wavevectors .................................24 Figure 9 - Structure of Staphylococcal nuclease using X-Ray crystallography ..........30 Figure 10 - Schematic of optics ...................................................................................33 Figure 11 - The filtration system viewed from above..................................................35 Figure 12 - Schematic diagram of internal furnace......................................................37 Figure 13 – Interior and exterior of aluminum can ......................................................38 Figure 14 - The typical setup of the Brookhaven Instruments DLS software .............40 Figure 15 - The optical table during a Dynamic Light Scattering run .........................41 Figure 16 - Viscosity as a function of temperature ......................................................44 Figure 17 - Raw data and our programmed fit for a single exponential at 24ºC …….45 Figure 18 - Raw data our programmed fit for a developed double-exponential at 44 ºC .............................................................................................................45 Figure 19 - Radius of SNase WT as a function of temperature ...................................46 Figure 20 - Evolution of the unscaled correlation functions at different times at 45 ºC.................................................................................................................47 Figure 21 - Evolution of the scaled correlation functions at different times at 45 ºC.................................................................................................................48 Figure 22 - Time development at 45 ºC of the "molten globule" ................................49 Figure 23 - Time development at 45 ºC of the "floppy chain" ....................................49 Figure 24 - Floppy chain size (the larger radius from a double-exponential) at different temperatures ......................................................................................51 4 Introduction Proteins and Cell Biology: From DNA to Protein The process by which a protein attains a specific three-dimensional ―activated‖ conformation is considered by some to be the last fundamental aspect of cell biology that is still incompletely understood. The recent success of the Human Genome Project emphasizes the fact that even though we understand the ―language‖ of biology as expressed in base pairs and the genetic code, this knowledge can only mean so much if we cannot interpret the architectural blueprint of structure that arises from this language. While we now have the entire DNA base sequence for humanity at our disposal, we are still unsure how or why a specific sequence of DNA nucleotides finally results in precisely arranged proteins in three dimensions. The process of the amino acids within a protein finding their particular stable structure is called protein folding. The mass movement of scientists now studying the theory of protein folding and contributing to databases of particular proteins is called proteinomics. Before focusing specifically on the protein and the problem of why it folds to a particular state, it is important to understand why this is fundamental to cell biology. The most fundamental principle of biology lies in the transcription of deoxyribonucleic acid (DNA). A DNA molecule consists of a long sequence of combinations of four nucleotides in a double helix. A specific nucleotide bonds to another analogous nucleotide, so that one side of the helix is enough to know the sequence of the other side. Adenine (A) always bonds to thymine (T); guanine (G) always bonds to cytosine (C). DNA can be thought of as a long string of four letters: A, C, T, and G. 5 Ribonucleic acid (RNA) comes in several different varieties, and its main function is to serve as a working template for transferring the information contained on a given segment of DNA into a specific string of amino acids. RNA polymerase ―unzips‖ the double-stranded helix of DNA, allowing the mRNA nucleotides (the same as DNA nucleotides except with uracil, U, replacing thymine, T) to form a complementary string of letters based on the same principle of base pairs within DNA. This mRNA sequence is refined by various methods so that some introns are snipped out and not part of the mature mRNA transcript.1 The entire process of converting a sequence of DNA to a usable sequence of RNA is called transcription. The mRNA is formed in the nucleus of the cell, but is free to move to other parts of the cell. Specifically, the ribosomes often lining the rough endoplasmic reticulum are an important site for the next stage of protein assembly. Within the cytoplasm, tRNA is a sort of molecular harness for specific amino acids. Each tRNA molecule has an anticodon of three nucleotides. On the surface of the ribosome, each tRNA anticodon binds to a specific codon in the mRNA sequence, which again consists of three nucleotides. Each codon therefore corresponds to a specific amino acid. It is interesting to note that there are conceivably 64 different combinations of codons (43 = 64), but there are only 20 different amino acids. Some codons act as a ―STOP‖ codon, so that the amino acid string ceases once these codons are reached. Nature has a built-in mechanism to avoid errors in that if the third nucleotide in a codon is changed for some reason, it will result in the same amino acid.2 The system by which specific codons of three nucleotides correspond to specific amino acids is called the genetic code. This is the Rosetta stone of cell biology, converting a long string of letters in mRNA into three-letter ―words‖ and 6 meaningful information in amino acids. The process by which an mRNA sequence is converted to an amino acid sequence is called translation. Amino acids are often called the building blocks of life, and we have now seen how the information contained in DNA eventually becomes a sequence of amino acids. Amino acids make up proteins, one of the most important and most diverse of all biological molecules. An amino acid is a small organic compound consisting of an amino group, a carboxyl group (an acid), a hydrogen atom, and one or more atoms known as an R group (see Figure 1). The R Group is what distinguishes the amino acids from each other. Figure 1 - Generalized structure of an amino acid. (From http://www.bact.wisc.edu/MicrotextBook/BacterialStructure/Proteins.html) When the tRNA brings each particular amino acid into proximity with the adjacent amino acids along the mRNA chain, these amino acids bond to each other. An amino group bonds to the adjacent carboxyl group, so that a regular pattern emerges: -NC-C-N-C-C-. This bond is called a peptide bond, and is extremely durable and stable (see Figure 2). The initial linkage of amino acids forms a polypeptide chain (distinct from the term protein because of a lack of three-dimensional structure). At the most basic level, each protein consists of a specific sequence of amino acids in its polypeptide chain, and this sequence is called the primary structure of the protein. The chemical synthesis of proteins ceases once peptide bonds are formed at the ribosomes, so the rest of the protein story has to do with the three-dimensional 7 Figure 2 - A typical peptide bond. (From http://www.bact.wisc.edu/MicrotextBook/BacterialStructure/Proteins.html) arrangement of this polypeptide chain. Proteins are large, diverse, and obviously extremely important to any form of life. They come in many different varieties: enzymes, structural proteins, transport proteins, nutritious proteins, hormones, glycoproteins, lipoproteins, and lymphatic proteins. Much of what we are comes from proteins which amazingly come from just twenty amino acids which come from just four DNA nucleotides. The key to this variety and to having a protein that is activated and can actually fulfill its intended purpose is the three-dimensional arrangement of the protein. To this point, the mechanisms governing the steps from DNA to protein are for the most part very well understood. After the primary structure of a protein, though, scientists are unsure about the exact mechanisms which govern a protein into a specific form in free space. Protein Structure A protein’s structure is the key to its use and function. As already stated, the primary structure of a protein is the unique sequence of amino acids in its polypeptide chain. This is the first level of structure and can be written on a page in linear fashion. The second level of structure is the coiled or extended shape of a protein. This is primarily due to hydrogen and disulfide bonds on adjacent links of a polypeptide chain. 8 It is not uncommon for hydrogen bonds to form between every third amino acid, for example.3 The peptide bond allows for a surprising amount of swivel and rotation so that covalent bonds can form with neighboring atoms. Sulfhydyrl linkages, or disulfide bonds, are covalent bonds between cysteine groups. Cysteine has a sulfur group available for binding to other groups and often forms a covalent link to another cysteine group within a protein (see Figure 3). Figure 3 - The structure of a sulfhydryl bond (disulfide bridge) between cysteine groups. (From http://www.bact.wisc.edu/MicrotextBook/BacterialStructure/Proteins.html) The secondary structure of a protein often results in two common forms: the αhelix and the β-sheet (sometimes called ―pleated,‖ see Figure 5). The formation of both arises from hydrogen bonds, disulfide bridges, and hydrophobic interactions of amino acids with water in solution. The α-helix resembles a ribbon of amino acids wrapped around a tube to form a staircase-like structure. The structure is very stable yet still flexible, allowing for other levels of structure. See Figure 4 for an example of an α-helix. In the β-sheet, two planes of amino acids form, lining up parallel so that hydrogen bonds form between each sheet. Unlike the α-helix, in the β-sheet hydrogen bonds form between amino acids very far away from each other within the primary structure. 4 Figure 5 shows an example of the β-sheet. Other variations on these general sorts of secondary structure are β-turns, random coils, helical wheels, terminal arms, and loops. 9 Figure 4 - A typical secondary structure of a protein, the ribbon α-helix. (From http://www.bact.wisc.edu/MicrotextBook/BacterialStructure/Proteins.html) Figure 5 - An example of the β -sheet, part of the secondary structure of a protein. (From http://www.bact.wisc.edu/MicrotextBook/BacterialStructure/Proteins.html) The tertiary structure consists of further folding of a coiled chain owing to bendproducing amino acids and interactions among R-groups far apart on a looped-out chain. Interactions between certain amino acids (such as proline) bend a chain at a certain length and at a certain angle. R-groups further down the length interact to hold the loop at characteristic positions. A tertiary structure can be viewed as a three-dimensional ―packing‖ of secondary structural elements. In many cases it is possible to identify recurring patterns of secondary structural organization, called motifs. 10 The tertiary structure is often most important in function. Proteins with very different primary structures but similar functions will often have similar tertiary structures. Figure 6 shows the tertiary structure for RubisCo, an enzyme used in the conversion of carbon dioxide to carbohydrates. Figure 6 - The tertiary structure of ribulose bisphosphate carboxylase (RubisCo), an important enzyme used in the conversion of carbon dioxide to carbohydrates. (From http://www.bact.wisc.edu/MicrotextBook/BacterialStructure/Proteins2.html) The final level of protein structure is the quaternary structure. In this level of interaction, two or more polypeptide chains are linked tightly by hydrogen bonds and other interactions such as hydrophobia, disulfide bridges, and R-group interaction. Inorganic cations can also play a role in this level of structure. Figure 7 shows a good example of the quaternary level of structure in hemoglobin, the oxygen-transporting protein in blood, which contains a heme group with an iron cation. 11 Figure 7 - The structure of hemoglobin, a good example of quaternary protein structure. (From http://www.isat.jmu.edu/users/klevicca/isat454/hemoglobin_essay.htm) The Problem With Folding The various levels of structure of a protein are explained through principles of interaction that are well understood by scientists, but I prefaced this discussion by saying that the problem of predicting the three-dimensional structure of proteins is the fundamental area of cell biology still eluding an understood framework. The problem lies mostly in the astronomically large numbers of different conformations (or specific states in free space) a protein may undertake. Even a small protein of only 150 amino acids (the size of the protein used in this project) each with two rotatable bonds and with three possible orientations each, would have between 4150 and 9150 possible conformations. If such a protein were able to sample one million of these conformations per second, it would take longer than the lifetime of the universe to find the ―correct‖ conformation.5 Understandably, scientists are having great difficulty predicting higher levels of protein structure given a primary sequence of amino acids. This is quite frustrating since the three-dimensional structure of a protein is what makes that protein 12 useful. All of the knowledge we have acquired concerning DNA and transcription and translation can only go so far. We are able to read the language of biology with the genetic code, but we are still unable to interpret this blueprint to build and manipulate proteins from scratch. The protein folding problem is so fundamental to biology that the benefits to understanding it would be so vast and broad as to be immeasurable. Engineering proteins to have a specific shape and function would go a long way toward controlling cancer, viral infections, and aging affects. Several ailments are already related to the ―misfolding‖ of proteins, where a particular protein simply does not fold correctly so that metabolic pathways or disease resistance methods are halted. These illnesses include: Alzheimer’s disease, Mad Cow disease (Bovine Spongiform Encephalopathy or BSE), Lou Gehrig’s disease (Amyotrophic Lateral Sclerosis or ALS), Creutzfeldt-Jakob disease, and Parkinson’s disease.6 Protein Denaturation The conformation of a protein when it is in its activated or functional state is also called the ―native state‖ of the protein. After synthesis at a ribosome, a protein typically ―finds‖ its specific natured state (out of the billions of billions of possible states) on the timescale of a few minutes. This is quite amazing, obviously, since scientific models at this point cannot even usually predict the final tertiary and quaternary states of a protein on any timescale. Since starting from the primary structure of a protein is quite impractical given the number of possible conformations, scientists usually start in analyzing a protein by how it unfolds or denatures from its natured state. 13 A protein has very specific pH, temperature, and pressure ranges where it is folded such that it is functional. When these values are varied slightly, the weaker bonds in the higher levels of structure are the first to break. In this way, the process by which a protein reaches its natured state can be studied by watching the way bonds break. Some theorists contend that bonds should break in the reverse order that they are formed. An important known exception to this is in proteins that contain the disulfide bond, which only fold properly in the presence of an oxidant.7 When a protein is denatured, it no longer functions as it is intended. An everyday example of denaturation is the protein lysozyme contained in uncooked chicken eggs. As the egg is cooked, the heat destroys the weaker bonds contributing to the three-dimensional structure, but does not disrupt the strong covalent bonds of the primary structure. This is what gives cooked eggs its texture and white foamy appearance. A major hurtle in denaturation is that much of the time when a protein is denatured, there is no way for the protein to fold back to its original state. There is no way to uncook an egg, and it is often not possible recover the long-range interactions that form the higher orders of protein structure. It seems to me that the hypothesis that a protein unfolds in the reverse order that it folds is not necessarily true for many proteins. For proteins that do refold to their natured state after denaturation, this hypothesis seems more feasible. For this project, Staphylococcal nuclease was chosen because it readily refolds to its natured state after denaturation and contains no disulfide bonds. The pathway a protein takes to its natured state is a hotly debated subject, with several plausible explanations available. Some think proteins nature all-at-once or at least into discernable stages corresponding to their order of structure. There is some 14 evidence for denaturation in stages, because proteins can form intermediate conformations called molten globules. It is likely some proteins fold rapidly and some less rapidly and in stages. Most agree that the Thermodynamic Hypothesis is critical in the formation of protein structure. This was proposed in the 1960s by one of the pioneers in protein analysis, C. B. Anfinsen.8 This hypothesis states that the native conformation of a protein is adopted spontaneously to ―the global minimum of [Gibbs] free energy.‖9 Gibbs free energy is a thermodynamic property used to determine whether a process is spontaneous or not. A system seeks out a minimum in Gibbs free energy according to the Second Law of Thermodynamics (entropy increases for any process). Further complicating the study of folding is the influence of nucleation, which is a specific event that triggers rapid folding.10 Methods of Protein Analyzation Common methods of analyzing the denaturing process include: Nuclear Magnetic Resonance spectroscopy (NMR), X-Ray crystallography, fluoroscopy and fluorescence correlation spectroscopy, and Dynamic Light Scattering (DLS). NMR spectroscopy utilizes the magnetic spin of the nuclei of atoms within a protein and aligns them with a strong external magnetic field. Nuclei with an odd number of protons or neutrons will have a net magnetic moment. This net magnetic field is then analyzed by pulsing radiofrequency (RF) radiation to excite the nuclei which will then emit specific frequencies of radiation depending on their position relative to other atoms in the protein. By using RF pulses at several angles, intensities, and frequencies, a three-dimensional picture of the protein can then emerge. A complicated process called sequential assignment is also used based on the amino acids known to be present. The 15 technique is quite time-consuming, and sequential assignment’s complexity means that NMR spectroscopy can only be used on small proteins (less than 200 amino acids in primary structure). Also, the environmental conditions of the protein must be held constant over long periods of time, lending high temperature and pressure studies impractical. Another newer method of NMR spectroscopy involves halting the protein folding process while it is occurring through a sudden change in pH and switching the solution to ―heavy water‖ or D2O. This ―paints‖ the protein at a specific point midfold so that each area of a protein can be marked as folding either before or after the process. This promising avenue does have its difficulty in controlling uniformly the point in the folding process at which the solution switch happens since the folding process is so rapid. Also, the sudden change in pH is enough to invasively disrupt the folding process and can alter the natural folding process. This data are also notoriously difficult to interpret. In X-Ray crystallography, proteins are first crystallized. This hinders the adaptability of the technique from the start since conditions must be very controlled and cannot vary. Crystallization is ―as much an art as a science . . . countless attempts to determine molecular structures have failed at this stage.‖11 Once a crystal is finally obtained, a diffraction pattern is produced by X-irradiation. This pattern consists of thousands of spots which are the raw data. The position and intensity of each spot is relatively easily determined, but the phases of the waves which formed each spot must also be determined in order to produce an electron density map. Once the phase problem is solved, a very accurate model of the protein’s structure can emerge, though the primary structure of the protein must already be known. X-Ray crystallography can be utilized 16 for any size protein, but does not give information about denatured proteins or even thermal conformational variability within a natured protein. In fluoroscopy, a few specific amino acids are excited by visible light and release electromagnetic radiation as they return to their ground states. The stereochemistry of the polypeptide chain and other environmental factors affect the fluoroscopy in ways that can be analyzed to follow changes in folding conformations. Fluoroscopy can give information about a protein’s conformation state, binding sites, solvent interactions, degrees of flexibility, internal motions, rotational diffusion coefficient, and other parameters.12 Fluoroscopy utilizes residues within a protein so that the fluorescence intensity from a solvent-exposed amino acid will be higher than that of the inner residue. This yields some information about how a protein unfolds by interpreting the changing amount of exposure to the residue and to the solvent. This method is somewhat crude compared to other methods and the residue may affect a protein’s folding, but fluoroscopy can provide a rough picture of the unfolding process. The technique used for this project is Dynamic Light Scattering (DLS), also known as Photon Correlation Spectroscopy (PCS). DLS does not yield as much structural information as NMR spectroscopy or X-Ray crystallography, but its strengths lie in versatility. NMR spectroscopy relies on a concentrated sample with very precisely managed environmental factors and can only study small proteins. X-Ray crystallography can only study proteins that are crystallized and therefore the amount of environmental factors affecting protein folding that can be studied is extremely limited. Also, both of these methods rely on very complex analysis. Fluoroscopy gives only a crude sense of protein unfolding and the residue and technique make the process 17 unnatural and invasive. DLS studies protein unfolding by shining coherent light upon a protein solution and analyzing the individual photons that are scattered off the protein. A detector transforms the signal from the individual photons into an electronic signal that can be analyzed by a computer quickly and without intrusion on the protein. The technique allows for great flexibility of the environment of the protein and does not involve any additives that might affect denaturation. The objective of DLS is to analyze the seemingly random properties of the scattered light intensity by calculating correlations in this time-dependent signal. When a coherent beam of light, such as a laser beam, passes through a solution, the solute particles scatter some of the light in all directions. When these particles are small compared to the wavelength of light used, the intensity of the scattered light is uniform in all directions. This is known as Rayleigh scattering. For larger particles, Mie scattering occurs where the intensity of the scattered light is angle- and wavelengthdependent. In this project, 514.5-nm coherent laser light is scattered on much smaller proteins or polystyrene spheres, and Rayleigh scattering occurs. In this project, coherent and monochromatic laser light is scattered by a solution of protein (or polystyrene spheres). A time-dependent fluctuation in the intensity of the scattered light at a particular angle is observed because the particles are small enough to undergo random thermal Brownian motion, so the distance between each protein molecule is constantly varying. At each instant, the light scattered from neighboring particles interferes either destructively or constructively so the Rayleigh scattering undergoes fluctuations in intensity. Since this fluctuation arises from properties of the solution, we can analyze the scattered light to yield information on the size of particles 18 within the solution. In particular, an analysis of the time-dependence of the intensity fluctuations yields the diffusion coefficient of the particles from which their hydrodynamic radius can be determined via the Stokes-Einstein equation. Thus, DLS is a quick and non-intrusive method of determining the size of a protein at a particular set of conditions. A photomultiplier tube (PMT) is used to detect single scattered photons within specific time intervals. A PMT consists of a series of plates so that a single photon is amplified into a cascade of electrons. Richard Feynman explains further how a PMT works: ―when a photon hits the metal plate A . . . , it causes an electron to break loose from one of the atoms in the plate. The free electron is strongly attracted to metal plate B (which has a positive charge on it) and hits it with enough force to break loose three or four electrons. Each of the electrons of plate B is attracted to plate C (which is also charged), and their collision with plate C knocks loose even more electrons. The process is repeated ten or twelve times, until billions of electrons, enough to make a sizable electron current, hit the last plate.‖13 This electric current is fed directly into a computer processor where a program can interpret the data. DLS therefore allows for precise and quick measurement of single photons by a PMT. DLS is extremely useful in protein folding studies of larger proteins and of proteins undergoing thermal, chemical, or pressure denaturation. The volume used can be small and relatively dilute and no foreign constituents need to be added to the solution for the process to occur. DLS is a method of actually watching the protein as it unfolds, and this voyeurism does not affect the folding or unfolding of the protein. DLS is 19 literally where physics and biology intersect, so that the mysteries of life can be studied according to the calculated objectivity of physical equations. Thermal Denaturation Dynamic Light Scattering can be used to study the particular effects on denaturation of almost any environmental stimulus. The goal of this project is to refine the DLS apparatus for thermal denaturation and to study the particular protein Staphylococcal nuclease. The focus of much of the rest of the discussion will thus be about temperature-dependence of protein denaturation, an area that is notoriously difficult to study with NMR spectroscopy and X-Ray diffraction. Thermal denaturation is important biologically because small changes in temperature can affect a protein’s solubility, enzymatic activity, and deactivate the protein, often irreversibly. For example, raising body temperature after the onset of a sickness activates certain enzymes within the immune system while raising the metabolism and helping to hinder proteins within the pathogens. An increase in temperature means that bonds within the protein molecule are strained and weakened. The weakest bonds are affected first and most severely. These are the bonds that form the tertiary (or quaternary, if present) levels of protein structure. The protein literally changes its shape as hydrogen bonds are broken and different areas of the molecule are exposed to the solvent. Water forms new hydrogen bonds with the amide nitrogen and carboxyl oxygen of the peptide bonds. Hydrophobic portions once on the interior of the folded protein are also exposed to the solvent. This increases the amount of water bound to each protein molecule. Thermal denaturation thus results in an increase in the hydrodynamic radius of the molecule affecting the viscosity and 20 solubility. The protein responds according to the Thermodynamic Hypothesis to minimize its Gibbs free energy by exposing as many polar groups and burying as many hydrophobic (literally ―water-fearing‖) groups as possible. This greatly affects internal polypeptide interaction so that the protein unfolds and gives a structure quite different than the activated protein at lower temperature. If the temperature returns to its original state, the protein may not return to its original structure even though the original state was likely lower in Gibbs free energy. Irreversible bonds, along with kinetic factors, may have formed so that the protein is unable to attain the necessary activation energy to return to its native state. The protein may only return to its natured state if all levels of structure except the primary one are broken. The hydrophobic groups of a protein contribute to an energy barrier that possibly inhibits a protein in returning to its natured state once the temperature returns to a normal range. If hydrophobic interactions are minimized and disulfide bridges are not present, a protein may return to its natured state after thermal denaturation (as is the case with Staphylococcal nuclease, the protein utilized in this project). Goals This research is part of a bigger project between several research groups to use Dynamic Light Scattering techniques to study protein folding in general and Staphylococcal nuclease and its mutants specifically. The primary goal of this thesis project involved refining the system apparatus for thermal denaturation and then using this apparatus to improve upon earlier incomplete thermal measurements on Staphylococcal nuclease. A new furnace design was successfully implemented that 21 allowed for the successful attainment of this goal as well as a study of the kinetics of unfolding in Staphylococcal nuclease. This study used 21-nm polystyrene spheres for test runs and to test the thermal DLS apparatus and Staphylococcal nuclease wild-type (abbreviated ―WT‖, the most common form of Staphylococcal nuclease) to obtain more reliable data about its thermal denaturation. Further projects are intended to study how substituting specific amino acids for Staphylococcal nuclease wild-type will affect thermal denaturation and other types of folding. Reliable high temperature data on Staphylococcal nuclease is not available, and this information will help in the development of this project and on the frontline of proteinomics in eventually being able to predict how a protein will fold under conditions given only the primary structure of a protein. 22 Theory of Dynamic Light Scattering Dynamic Light Scattering in its fundamental form consists of shining coherent electromagnetic radiation upon a solution, collecting the scattered photons at a certain angle, and using this data to uncover characteristics of the solution. The technique is not limited to biological systems, though it is fast becoming a convenient tool for medical diagnostics. For example, DLS has been proven to be extremely effective in the very early detection of cataracts in the eye, an ailment responsible for half the cases of blindness and prevalent in 34 million people aged over 65.14 Other applications of DLS include nanoscale engineering, feedback systems for water filtration, diagnostics for the purity of solutions, and general characteristics of solutions. Its main use is the noninvasive determination of solute particle size in the dilute limit of a solution, which of course lends itself well to protein denaturation research. When a coherent (in phase) laser beam illuminates a solution, the distribution of charges within the molecules in solution is subjected to an oscillating electric field, and the charges are therefore accelerated. Classical electromagnetic theory proves that an accelerated charge will radiate electromagnetic radiation. If the solution is optically similar (its dielectric constant is homogeneous throughout the solution), the light radiated from each region of the solution differs only by a phase factor (i.e., the wavelength of light will be the same, but there will be patterns where the light destructively or constructively interferes). If the dielectric constant fluctuates within the medium, as happens when molecules move within a solution (e.g. Brownian and thermal motion), light will be scattered in all directions and undergo frequency shifts. 23 DLS depends primarily on quasi-elastic scattering, which deals with light scattered from the translational and rotational degrees of freedom of solutes in a solution. ―Quasi-elastic‖ means that the scattered light has almost the same wavelength as the incident light.15 When frequencies are separated by a very small amount, this gives rise to a beating of light where the intensity seems to wax and wane with time. This beating of light can be detected by the PMT. In DLS, coherent incident laser light of wavevector i (the incident beam) passes through the scattering volume (the protein sample). Light with wavevector s (the scattered beam) is scattered by this sample and measured by the PMT at some known angle θ (in our case 90°) with respect to the exiting incident beam (see Figure 8 below). Figure 8 - Diagram of DLS incident and scattered wavevectors. For this project θ=90º. The incident beam is scattered in all directions, and the bottom arrow emphasizes this. The PMT transforms intensity fluctuations from the scattered light into a corresponding fluctuating voltage as described before. A computer records the output and calculates from it an intensity-intensity autocorrelation function. The computer can also analyze these correlation functions to yield physical data about the solution. 24 If the scatterer in solution moves (such as the Brownian and thermal motion for proteins), the dielectric constant fluctuates over time and a correlation function can be used. A correlation function is a way to interpret seemingly random data. In this application, light intensities defined to be similar to each other over a time interval are said to be correlated, and those that are not are considered not to be correlated. The following derivation of how the scattered light intensity that the PMT measures can correspond to the radius of particles in solution is normally calculated through Brookhaven Instruments DLS control software or on a mathematics program such as Mathcad or SigmaPlot. From advanced electromagnetic scattering theory, the intensity of scattered light from one molecule is 4 2 M 2 sin 2 ( dn / dc) 2 I 0 I (1) , N A2 4 R 2 (1) where M is the molecular weight of the molecule, υ is the angle of the scattered beam with respect to the polarization of the incident beam, R is the distance from the scattering medium to the detector, (dn/dc) is the rate of change of the index of refraction as the concentration of the solution changes, I0 is the intensity of the incident beam, NA is Avogadro’s number, and λ is the wavelength of light in the solution.16 To interpret light scattered by the solution, we need to calculate the intensity due to a large number of molecules. We need to calculate the electric field due to each molecule, and the total intensity will be proportional to the square of the sum of each individual electric field. The average intensity of a light beam given in terms of its maximum electric field is 25 2 E max I in W/m2. 2(3.77 ) (2) Assuming all scattering molecules are identical, so that electric fields vary in phase but not magnitude, the scattered electric field that is in phase with some reference beam is E si E s cos i , (3) i while the electric field out of phase with the same beam is E so E s sin i , (4) i where Es is the maximum magnitude of each scattered electric field, the sum is over the number of scattering molecules, and σi is the phase of the scattered field due to the i-th particle. Combining these equations gives 2 2 N 1 N 2 I (N ) E s cos i sin i , 2(3.77 ) i 1 i 1 (5) which, through trigonometry and substitution of Equations (1) and (2), becomes N I ( N ) I (1) N 2 (cos( i j ) , j i 1 (6) where N is the number of scattering particles. The terms that vary as the molecules move in solution are the σi’s and σj’s. The second term in Equation (6) will vary over time and so will the net intensity of scattered light.17 This time of fluctuation between maximum and minimum intensity is strictly dependent on the properties of the scattering molecule, and this is the key to understanding how Dynamic Light Scattering works. By measuring the macroscopic fluctuations in the intensity of light scattered by a solution, we can determine microscopic properties of the solution. 26 Intensity fluctuations in DLS are usually studied via correlation functions. Consider an arbitrary function F(t). The time autocorrelation function of F is defined as: 1 T T F (0), F ( ) lim T 0 F (t ) F (t )dt . (7) This function measures how correlated the function F is at time τ compared to F at time 0. It is useful to convert the integral to a discrete small interval Δt, where F changes negligibly, t=iΔt, τ=jΔt, and T=NΔt, so Equation (7) becomes F (0), F ( ) lim N 1 N N F F i i j , (8) i 1 where Fi is the value of F at the start of the i-th interval and N is the number of molecules. In the case of DLS, the function is the number of photons collected by the PMT, Δt is called the sample time and τ is called the delay time. The number of photons incident on the tube during the i-th interval of Δt duration is ni, and the autocorrelation function is then defined as18 1 N N G ( 2 ) ( jt ) lim N n n i i j ni ni j . (9) i 1 The intensity autocorrelation function is the signal measured during DLS experiments, but the properties of the scattered molecules are more easily calculated from the electric field autocorrelation function. To obtain this, we need to normalize the intensity autocorrelation function with respect to a baseline value to get the net intensity autocorrelation function: g ( 2 ) (t ) G ( 2 ) (t ) , G ( 2 ) ( ) 27 (10) where G(2)(∞) is the baseline signal measured over a long period of time. This is converted to the Electric Field Autocorrelation function g(1)(t) using the Siegert Relationship19: 2 g ( 2) (t ) 1 g (1) (t ) , (11) where 0<β<1 is dependent on experimental values (e.g. properties of the laser beam, detection optics) and is calculated during data fitting. The electric field autocorrelation function is determined by Fourier transform of the number density of the scattered light20: g 1 (r , t ) nˆ d (k , t ), nˆ d (k ,0) . (12) The Siegert relationship is based on the assumption that the scattered electric field has a Gaussian distribution and is valid whenever the system is ergodic, when the timeaveraged properties of the system are equal to the ensemble-averaged properties. Computer algorithms often help in finding a functional fit for g(1)(t) since it is notoriously difficult to calculate in general. Consider the case for a dilute solution of rigid, monodisperse spheres undergoing only translational diffusion, the function g(1)(t) is simply g (1) (t ) e 2 t , (13) where Γ is called the decay rate, which is related to the translational diffusion constant DT of the molecule by DT q 2 , (14) where q is the scattering vector defined as the difference between the scattered ( s ) and incident ( i ) wavevectors (see again Figure 8): q k s ki , 28 (15) where the magnitude of ks = (2/s) and of ki = (2/i). The magnitude of q is q 4 n 0 sin , 2 (16) where n is the index of refraction of the solution, 0 is the wavelength of the incident light in vacuum, and is the scattering angle inside the solution. In the dilute limit, particle interactions can be ignored, and the diffusion constant (DT) for spheres in a solution with bulk viscosity η is given by the Stokes-Einstein Equation: DT k BT , 6rh (17) where kB is Boltzmann’s constant, T is the temperature of the solution in Kelvin, and rh is the hydrodynamic radius of the scattering molecule. The autocorrelation function has thus come full circle to tell us information about the size of the scatterer with the assumption that they are spheres. We will use this assumption that in its native state, a protein is roughly a sphere, and that the radius of the sphere increases as the sphere unfolds. Equation (13) is also valid in the case where several different size spheres are present. This case gives a correlation function with a double decay exponential instead of a single decay exponential. When two or more particles are in a solution, we can extract the spherical radius of these particles by analyzing the correlation function as a sum of two or more decaying exponentials. Staphylococcal nuclease is globular in its native state, so the spherical assumption does yield meaningful data about its size. 29 Properties of Staphylococcal nuclease Staphylococcal nuclease (sometimes abbreviated ―SNase‖) is a DNA and RNA cleaving enzyme with the property of transition between its denatured and natured states along with, stability under relatively extreme conditions. This makes SNase ideal for a DLS study. SNase contains no disulfide bridges, so it is possible to study the protein’s folding process by studying its denaturation in reverse. Staphylococcal nuclease consists of 149 amino acids, has no quaternary structure, and has molecular mass of 16.8 kDa.21 Shortle & Ackerman have recorded its hydrodynamic radius as 1.6 nm in its natured state and 3.5 nm when denatured.22 SNase can withstand temperatures as high as 65 ºC and is stable in the presence of other enzymes. This project utilizes the naturally occurring ―wild-type‖ SNase (WT); future projects may use mutants. Figure 9 below shows the structure of Staphylococcal nuclease. Figure 9 - Structure of Staphylococcal nuclease using X-Ray crystallography. (From http://www.rcsb.org/pdb/) 30 Experimental Method Production of Staphylococcal nuclease The production of Staphylococcal nuclease wild-type (WT) is fairly timeconsuming, taking about twenty steps over the course of four or five days. The following will be only a brief qualitative summary of how Staphylococcal nuclease is produced from scratch in a biochemistry lab. For a more in-depth discussion on the step-by-step production of WT, see April Fortner’s 2002 University of Arkansas Honors Thesis entitled: ―Protein Chemical Denaturation and Analysis using Dynamic Light Scattering,‖23 and supplemented with Dr. Wes Stites’ laboratory protocol for protein production: ―Nuclease Protein Preparation in λ Expression System.‖24 An ampicillin-resistant strain of E. coli containing a gene for Staphylococcal nuclease expression is first grown in a broth and agitated at a warm temperature to produce an abundance of protein. The bacterial E. coli cells are then lysed, so that protein is released into solution. Ampicillin is added, which prevents other bacteria from being produced. Care is taken not to expose the media to air, so that outside contaminants are not introduced. Once the bacterial cells are lysed, it simply becomes a laborious process of isolating the WT from the many other contents inside and outside of the lysed cell. This is achieved through a long series of centrifugations, resuspensions in buffer, and cation exchange columns. The initial step of the procedure typically begins with about 3 L of broth containing the bacterial sample and by the end of the process, a yield of about 60 mg of WT in a buffer solution is the reward for the lengthy torture. 31 Dynamic Light Scattering Procedure Several critical points need to be kept in mind when undertaking the procedure for analyzing a protein sample by Dynamic Light Scattering. First, an optical axis needs to be carefully established so that the incident and scattered angles are accurately known. Second, all care must be taken to keep the sample as pure as possible so that scattering from other solutes such as dust and contaminants contained within the solution is kept to a negligible level. Third, it is imperative that the only photons reaching the photomultiplier tube are the photons scattered off the sample solution. This means that any stray lights need to be off, the glass tubing for the sample needs to be extremely clean so that stray laser light does not scatter from it, and a method needs to be established of allowing only the photons scattered along the optical axis into the PMT. After all of this preparation, a clean signal can reach the PMT, which is converted to an electronic signal and analyzed by a computer. The procedure to follow is critical for fulfilling the above criteria. The optical setup is shown in Figure 10, with a Coherent Innova 306 laser operating in single mode at a wavelength of 514.5 nm supplying the coherent light. Since polystyrene spheres and SNase are much less than this wavelength, Rayleigh scattering will be observed. From the laser, the beam is split into the incident and alignment beams by a 90/10 beam splitter. A system of mirrors brings the two beams roughly perpendicular, and the angle of 90 degrees is finally established by mounting a mirror where the sample would normally be and orienting it to 45º 0’ (± 5’). The incident beam is directed onto this mirror such that it counter-propagates along the alignment beam so that after removing the mirror, the incident beam is at an angle of 90.0º (± .1º) with 32 respect to the alignment beam on the axis with the PMT. Once this angle is established, there is no need to repeat this procedure each time unless the table or mirrors are bumped significantly. Figure 10 - Schematic of optics. M=mirror, L=lens. Lens L1 (focal length=63.5 mm) is used to focus the incident beam onto the sample. This focuses the radius of the beam in the focal plane to as little as 5 μm. Lens L2 (focal length=100 mm) focuses the scattered light onto the 0.05 mm pinhole, which, along with the iris, helps to reduce the amount of unwanted light that enters the PMT and establish a single coherence area over the size of the PMT’s photocathode. During experiments, a black light-tight box is placed over the PMT and black fabric drapes the cracks to ensure further protection from unwanted photons. Lens L2 was assembled in a 2f-2f configuration meaning that there is a distance of 200 mm (twice the focal length) 33 from the scattering volume to the lens as well as from the lens and to the pinhole. This means that the magnification is one since the object and image distances are equal. A microscope was used to aid in finding the scattered beam and for ensuring that light was collected along a diameter of the cylindrical glass sample chamber to minimize error due to refraction. The PMT is connected to a Brookhaven Instruments BI9000 digital autocorrelator board in a standard PC. A Thorne-EMI PMT is used with a transit time of less than 25 ns, which means the smallest delay time possible is 100 ns. The voltage applied to the PMT was usually between -1.9 and -2.2 kV giving 25 to 50 kilocounts per second upon excitation by scattered light. This voltage was tuned to give a stable count rate. The count rate was also regulated by neutral density filters on the optical table, which filter the incident laser beam. Another piece of information that needs to be kept in mind when performing a DLS experiment is the idea of coherence area. The section of the interference pattern detected by the photocathode of the PMT needs to be small enough that solitary fluctuations in intensity are observed, rather than intensity averages over many fluctuating points. Coherence area is defined as: Acoh 2 R 2 , a 2 (18) where λ is the wavelength of the laser light, R is the distance from the scattering volume to the detector (in this case from the scattering volume on the pinhole to the detector, about 22 cm), and a is the radius of the scattering volume (at most 25 μm). This means the coherence area is about 2.6 mm. The area viewed by the detector should be less than 34 this value if possible, and certainly not much more. The size of the photocathode is about 5 mm in diameter, so this requirement is filled. Of utmost importance is a clean sample and this is achieved by filtering the sample and cleaning the constituent parts of this system. Valves and glassware are cleaned with a solution of filtered water and Alconox cleaner. The valves and glassware are then placed in a sonicator for fifteen minutes and cleansed again in deionized water. The filtration apparatus consists of various Upchurch Scientific valves, two 0.2 μm inline filters, a thin glass tube where the laser is focused on the sample (outer diameter of 1/8 inch and inner diameter of 3/32 inch). A glass blower heated the ends of the tubing until their end diameters created a tight seal (sometimes frustratingly too tight) with the 1/16 outer diameter Teflon tubing used to connect various parts of the filter. The pump, a BISFS model from Brookhaven Instruments Corporation, controls the flowrate of the filtration. Before taking data, the system was filtered for at least twenty minutes to ensure a clean sample. Once filtered, the valves are closed so that liquid cannot leave or enter the glass tube. See Figure 11 for an overview of the filtration system. Figure 11 - The filtration system viewed from above. 35 The heating apparatus consists of an internal furnace where the glass tube holding the sample is inserted along with an aluminum capped cylinder can to isolate the sample from room temperature air. Prior attempts at thermal denaturation were plagued by a lack of stable rate counts at higher temperatures probably due to convection currents, and the aluminum cylinder encasing the internal heating device eliminated these convection currents by ensuring that the entire sample contained between closed valves was held at the same temperature. The internal furnace used to be exposed to room temperature air, and besides the troublesome convection currents, it was not as successful at holding a stable temperature as the new cylinder encasing system is. The internal heater’s design (see Figure 12) consists of two pieces of circular aluminum 1.5 cm in height and 2.2 cm in diameter both of which have 0.2-cm holes drilled through their centers. The two pieces are separated by spacers 0.5 cm in length so that there is a 0.5 cm gap between the top and bottom of the furnace, so both the incident and scattered beams can enter. Minco heaters were wrapped around the outside of the aluminum pieces and a thermocouple was placed along the bottom heater and wrapped in insulation tape. A thermocouple is placed along the central cavity which measures the temperature against the glass tube holding the sample. The heaters and thermocouples are connected to a Lake Shore 330 Autotuning Temperature Control Unit allowing for a digital readout of the temperature at each thermocouple, along with heating control to the thermocouple near the sample. 36 Figure 12 - Schematic diagram of internal furnace. The aluminum can encasing the internal heater that Dr. Oliver and I designed is constructed to be form-fitted on the bottom and top lids to the valves. This ensures that the entire sample solution is contained within an environment of the same temperature when the valves are closed. Before, the internal heater was too small to encase the entire sample tube and valves. The internal furnace is now mounted on the inside of the can so that it extends along the vertical element of the entire can. The can is 4.5 inches in height and 2.5 inches in diameter. Half of the cylinder ―shell‖ can be removed allowing for easy access to the internal part of the can for alignment purposes. Four windows were constructed ninety degrees apart, so that the incident, scattered, and exiting beams are all allowed passage out. The fourth window is used for wires to exit the can. Once the beams and sample are roughly aligned, the other half of the cylindrical shell can be placed back on to insulate the sample from the room temperature air. Heating tape (of 37 80-Ω resistance) is then wrapped around the can making sure that none of the windows (besides the wire window) are covered. The heating tape is connected to a 0-130 V Tenma Variable Auto Transformer, which is used for rough temperature control. Once the temperature is near what is desired, the internal heating control can ensure minute control (along with a digital read-out). This temperature controlling system allowed for little fluctuation (usually less than 0.03 °C) and could be held at the same temperature for any desired length of time. Figure 13 shows the aluminum can with the shell exposing the interior and also the can wrapped in heating tape, as it appeared during a trial run. Figure 13 - Left: aluminum can with shell removed revealing the interior heater, thermocouples, and the glass tube holding the sample. Right: the aluminum can with heating tape wrapped around it and insulation. A Staphylococcal nuclease sample is prepared by first diluting the sample to about 10 mg/mL and thawing it in a 55 ºC water bath for five minutes to remove dimers (bonding between protein molecules during the freezing process). This solution is then diluted further with the 0.1 M NaCl, 0.025 M NaPO4 buffer solution. The physical equations described before are valid in dilute limits, so the more dilute the sample that 38 can be obtained that still gives a strong scattered signal, the better. A final concentration of about 7 mg of protein per milliliter of solution (about 0.6% by mass) was typical and at least 5 mL of sample is needed. This solution is introduced into the filtration system (usually for about half an hour), the can apparatus and heating tape is then installed, the alignment is checked again, the sample is heated up using the temperature controlling unit (for fine adjustments and feedback control) and Variable Auto Transformer (for coarse adjustments), and data is finally taken. 21-nm polystyrene sphere solutions are prepared by simply adding 3 to 5 drops of Duke Scientific polystyrene sphere solution (density 1.05 g/mL) to about 15 mL of water or buffer solution and similarly setting up the system apparatus. Brookhaven Instruments Dynamic Light Scattering Software records and interprets data instantaneously as the apparatus is running. Neutral density filters on the optics table and the voltage are tuned so that at room temperature, the count rate from the PMT is stable and somewhat low (25 to 35 kcps). Any adjustment of the voltage will correspond to a change in count rate which the software will interpret as a change emanating from the sample. For this reason, it is undesirable to change the voltage, and if this does occur the run must be started over. Since count rates typically increase with increases in temperature, the count rate gradually increases to around 50 kcps at temperatures of about 40 °C, so a voltage adjustment is unnecessary if it is set lower at low temperatures. At higher temperatures, it is somewhat difficult to avoid the need to change voltage or the neutral density filters. High temperatures also give rise to a dynamic environment that makes it difficult to obtain a steady count rate. This will be explained more thoroughly in the Results & Discussion section. 39 The software has many functions that can be utilized, but runs typically consisted of the four windows shown in Figure 14: the Control window, NNLS (Non-Negatively Least Constrained Multipass), Count Rate History, and the Correlation Function. Figure 14 - The typical setup of the Brookhaven Instruments Dynamic Light Scattering software. Windows, clockwise from top left: Control Window, NNLS, Count Rate History, and Correlation Function Window. This run is for a 21-nm sphere solution. Notice the smooth correlation function, steady count rate, and the NNLS distribution of sphere size around 20 nm. The Control Window gives information about the delay times, the elapsed run time, the sample rate, the baseline, the baseline differential (less than 5% for meaningful data and typically under 1%). It also controls vital settings such as temperature, viscosity, the wavelength of light, and channel input, among others. NNLS is one of several standard methods the Brookhaven software uses to interpret the data as a distribution of particle sizes. It utilizes a complicated algorithm to give the particle sizes 40 while the computer is still collecting data. For the small proteins and polystyrene spheres used in this experiment, NNLS was found to be the most accurate given our set of conditions.25 The software gives a distribution of particle sizes, but we typically used the raw data and interpreted it through programs Dr. Oliver and I created in Mathcad since the software gives little user control and its built-in viscosity model assumes a dilute environment which did not match our system. After the lengthy preparation described in this experimental method, the optical table appears as in Figure 15 and data can be taken. Figure 15 - The optical table during a Dynamic Light Scattering run after all of the experimental method has been followed. 41 Results and Discussion The Theory of Dynamic Light Scattering section showed how a correlation function can be interpreted to use the Stokes-Einstein equation below by extracting the diffusion constant and the hydrodynamic radius (rk) assuming spherical scatterers: DT k BT , 6rh (19) where T is the temperature in Kelvin, kB is Boltzmann’s constant, and η is the bulk viscosity. The diffusion constant (DT) is determined by fitting an exponential to the correlation function as described before. Solving Equation (19) for the hydrodynamic radius gives rk k BT . 6DT (20) It is critical to have accurate temperature-dependent viscosity data in order to determine accurately the hydrodynamic radii of the solute particles. Unfortunately, viscosity varies greatly depending on slight changes in many factors. Relating to this project, viscosity changes with temperature, concentration of chemicals within solution, particle interactions, and electrostatic interactions between the solution and its container. The buffer solution we use is somewhat dilute, consisting of 0.1 M NaCl and 0.025 M NaPO4. Nadeem Akbar took viscosity data as a function of temperature of this buffer solution by using 21-nm spheres and treating this hydrodynamic radius as constant at any temperature. He was therefore able to extract the buffer solution’s viscosity in much the same way that we extract particle size. The viscosity data he took was somewhat lower 42 than water for all temperatures. This does not make much physical sense because dissolution of ions in water (including NaCl and NaPO4) nearly always raises, not lowers, the viscosity of the solution compared to deionized water.26 Several factors could have given skewed data including an inflation of size of the spheres at increasing temperatures, electrostatic interactions between the spheres, glass, and water, and a solution too concentrated in spheres. The protein solutions used in this project were typically about 7 mg/mL— much more concentrated than the buffer solution. This viscosity data is therefore not applicable to a solution containing large organic molecules, which would likely raise the viscosity even more than the buffer solution. Given that we could not assume the size of proteins, we assumed that the Akbar viscosity data held its trend with temperature even if its values were problematic. Near room temperature, Staphylococcal nuclease is tightly folded and our data yielded consistent sizes no matter which viscosity value is used. We decided to scale the Akbar viscosity data so that at room temperatures, our protein size agreed with Shortle and Ackerman’s data of about 1.6 nm in radius for SNase in its natured state. The scaled Akbar viscosity data gives viscosity as a function of temperature and yields the agreed-upon value for the natured state’s size. Figure 16 shows a graph of the original data and the scaled data, which is the function we utilized. 43 Viscosity vs. T for Buffer 1.6 1.4 Temp (C) vs Visc (cP) Quadratic Fit Scaled Fit Viscosity (cPoise) 1.2 1.0 0.8 0.6 0.4 0.2 0.0 10 20 30 40 50 60 70 80 Temperature (°C) Figure 16 - Viscosity as a function of temperature. The Akbar data is in red and the scaled data we used is in blue. The NNLS window of the Brookhaven software gives a distribution of particle radii, but we found the software to be somewhat limiting in control, so we used Mathcad to analyze the raw data ourselves. NNLS used aqueous viscosities so the radii it reported were off, but its program gave us a good qualitative and even quantitative idea of error distributions. Temperatures near room temperature (about 23 ºC) all the way up to over 40 ºC gave fairly typical and clean single-exponential decaying correlation functions. The NNLS program likewise consistently interpreted the data to be a tight distribution of a very small particle all the way up to 40 ºC. Just before 45 ºC, the correlation function showed signs of developing a double-exponential—another, longer time decay. Short time decays mean that the scattering particle is small and longer time decays mean that the scattering particle is larger. Thus, at lower temperatures, we needed only to interpret single-exponential decaying functions, while at higher temperatures, a doubleexponential needed to be analyzed. See Figures 17 and 18 for examples. 44 Correlation Function and Fits at 24 ºC 1.45 1.40 Raw C() Data Single-Exponential Fit C() 1.35 1.30 1.25 1.20 1.15 100 101 102 103 104 105 106 107 Time in s Figure 17 - Raw data and our programmed fit for a single-exponential at 24 ºC. The radius is calculated to be 1.584 nm for this example. Correlation Function and Fits at 45 ºC 5.4 5.2 C() Raw C() Data Double-Exponential Fit 5.0 4.8 4.6 100 101 102 103 104 105 106 107 Time in s Figure 18 - Raw data and our programmed fit for a developed double-exponential at 45 ºC. The two radii were calculated to be 2.836 nm and 84.052 nm for this example. 45 Our program allowed us to use either a single-exponential fit or a doubleexponential fit. We also had the ability to choose which data points to fit by hand, because often at high temperatures, the baseline would have a downward slope due to thermal factors that were not meaningful to us, namely the count rate generally increases from the beginning to the end of a particular run at high temperatures. Also, the first few data points sometimes were scattered and we could sometimes do a better job of fitting than the computer program. We carefully analyzed the hydrodynamic radius as a function of temperature for Staphylococcal nuclease from 23 ºC up to 60 ºC. The enhanced temperature control and insulating system made it possible to take meaningful data all the way up to this temperature, whereas the prior system was unable to remain stable above 40 ºC. Figure 19 shows the radius (or smallest radius when double-exponentials arose) for Staphylococcal nuclease wild-type as a function of temperature. Radius (nm) SNase Radius vs. Temperature 6.0 5.5 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 20 25 30 35 40 45 50 55 60 Temperature (C) Figure 19 - Hydrodynamic radius of SNase WT as a function of temperature. For higher temperatures (>43 ºC) the radius plotted is the smaller of the two radii in double-exponential fits. 46 The graph in Figure 19 clearly shows a gradual increase of the radius from 1.6 nm to 2.1 nm until just before 45 ºC. The radius rapidly becomes larger at this point until it settles just under 5.0 nm. The system at 45 ºC deserves special attention. As already mentioned, doubleexponentials begin forming just before this temperature. This means that other larger components emerge at around this temperature that were not in the system before. The temperature was held constant at 45 ºC and the correlation function was allowed to evolve to further study the kinetic emergence of large particles. Figures 20 and 21 show the evolution over time for the correlation function at 45 ºC (scaled and unscaled). Evolution of the Raw SNase Correlation Function at 45 ºC 10x10 6 8x106 C() after 0-2 mins C() after 3-5 mins C() after 6-8 mins C() after 8-10 mins C() after 10-12 mins C() 6x106 4x106 2x106 0 100 101 102 103 104 105 106 107 Time in s Figure 20 - Evolution of the unscaled correlation functions at different times at 45 ºC. The background evolution is evidence of an increase in count rate, leveling off after about 12 minutes. 47 Evolution of the Scaled SNase Correlation Function at 45 ºC 1.5 1.0 C() after 0-2 mins C() after 3-5 mins C() after 6-8 mins C() after 8-10 mins C() after 10-12 mins C() X + +++++ + 0.5 + XXXXXX + + X X 0.0 -0.5 100 ++ + X X +++++++++++++ +++++++++++++++++++++++++++++ XXXXXXXXXXXXXX +X +X +XX +XX +X +X +X +X +X +XX +X +X +X X XXX +X +X +X +X +XX +X +XX +X X +X +X +X +X +X X X +X +X +X +X +X X X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X X X X X X XX X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X XXX X X X +X +X +X +X +X +X +X X X X +X +X +X +X +X +X X X XX +X +X +X +X +X +X + +X +X +X +X +X X X XX +X +X +X +X +X X +X +XX +X +X +X +XX +X +X +X +X +X +X +X X +XX +X +XX +X +X +X +XX +XX +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X +X + 101 102 103 104 105 106 X + 107 Time in s Figure 21 – Evolution of the scaled correlation functions at different times at 45 °C. Notice how the double-exponential develops by 3 minutes and matures by 12 minutes. Our data fits for this time evolution at 45 ºC yield a very developed doubleexponential (to a reasonable approximation) after twelve minutes as well as an increasing smaller radius which I will refer to as a molten globule. Figures 22 and 23 graph this data. Gast et al. refer to molten globules as ―nearly as compact as the native state, have native-like secondary structure, and differ mainly by the lack of a rigid tertiary structure.‖27 The molten globule can be thought of as a swelling protein that retains some structure, though the tertiary and secondary structures are dynamically changing, giving a bigger average radius. I refer to the bigger radius obtained from the double-exponential curves as a ―floppy chain.‖ Since the larger particles were not in the system below 45 ºC, I conclude that the larger particles that develop are indeed denatured polypeptide chains of SNase. Our analysis yielded floppy chains of about 80 nm depending on 48 radius (nm) SNase molten globule growth over time at 45 ºC 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 0 5 10 15 time (min) Figure 22 - Time development at 45 ºC of the "molten globule," the smaller of the radii obtained from the correlation function. SNase floppy chain growth over time at 45 ºC radius (nm) 100 75 50 25 0 0 5 10 15 time (min) Figure 23 - Time development at 45 ºC of the "floppy chain," the larger of the radii obtained from the correlation function. 49 the temperature and time evolution. The NNLS program likewise indicated an emergence of larger particles at high temperatures. Unlike the smaller molten globule radius, the larger particle had a wide radius distribution (usually about 20 nm), and yielded an average larger particle size smaller than our own analysis. Like our data, NNLS interpreted the floppy chains as becoming larger and a greater constituent of the solution with increases in temperature and developing over time. Given this information, we can conclude that large floppy chains do indeed form at higher temperatures and as the system sits at a high temperature, this effect evolves to become more pronounced. Since we assumed that proteins are approximated as spheres, it makes sense that long chains of polypeptides entangling with each other in a dynamic environment would yield a correlation function that consists of an obvious mixture of big particles. The correlation function of floppy chains would be unable to accurately obtain a spherical size for them because floppy chains are not spheres and they are constantly folding in on themselves and interacting with other chains. SNase has 149 amino acids, so a solution consisting of a turbulent mixture of chains lacking secondary structure and interacting with other chains could yield a fairly large scattering particle (100 nm). Figure 24 shows floppy chain size as a function of temperature. The data at 45 ºC is the same time development from Figure 23. The general trend that floppy chains become bigger and more pronounced at higher temperatures and after time evolution is valid even if we cannot interpret the floppy chains as having a specific spherical radius. 50 SNase floppy chain size Vs. Temperature 250 Radius (nm) 200 150 100 50 0 40 45 50 55 60 Temperature (ºC) Figure 24 - Floppy chain size (the larger radius from a double-exponential) at different temperatures. The general trend of larger floppy chains over time at a given high temperature and as temperature increases is valid even if the calculated sizes of the floppy chains are inaccurate since they are not spheres and the system is turbulent. The data points at 45 °C are the same from the time evolution experiments shown in Figures 20-23. 51 Conclusions This study was successful in analyzing the protein denaturation of Staphylococcal nuclease wild-type at high temperatures. The redesign of the heating and insulation system along with improved filtration and strict cleaning were all instrumental in allowing the system to remain stable at high temperatures and to be able to record meaningful data. Especially provocative is the emergence of floppy chains at high temperatures. Prior attempts to thermally denature SNase could only maintain a stable system up to about 40 ºC, so the phenomenon of floppy chains could not be accurately studied, only speculated. The floppy chains took about five minutes to form at 45 ºC and continued evolving for another ten minutes. This time evolution is an exciting prospect in protein studies because time evolution studies usually utilize chemical denaturation and the evolution stops usually after a minute. Thermal denaturation time evolution studies have not been performed yet to a great extent, and the emergence of floppy chains at times much greater than one minute is exciting. For SNase, these floppy chains refold into natured proteins after awhile and after the temperature is lowered. Even at temperatures lower than 45 ºC, the floppy chains remain when lowering the temperature from a high temperature, but completely refold to the native state after a few minutes. The results of this project raise more questions than they answer, interesting as the results might be. A continuation of this project would first need to obtain more reliable viscosity data for each specific protein solution that is studied. An interesting study would be the thermal refolding of the floppy chains as temperature is slowly lowered 52 from a temperature greater than 45 ºC. Thermal, chemical, and pressure or denaturation or combinations of those, along with mutants of Staphylococcal nuclease would also yield meaningful data. Analysis of thermal denaturation of single protein molecules using optical tweezers is also an exciting prospect for the continuation of this project. The system apparatus for studying protein denaturation through Dynamic Light Scattering has been greatly improved, allowing for continued studies and more extensive collaborations between the research laboratories of Dr. Oliver of the Physics Department and Dr. Stites of the Chemistry/Biochemistry department at the University of Arkansas. Frontline research in Proteinomics helps in our understanding of how proteins fold and this knowledge may someday result in being able to predict the final folded state of a protein given its primary sequence. Manipulation of protein structure and full understanding of the principles of cell biology could benefit humanity almost immeasurably by potentially eliminating diseases, and engineering bionanoparticles, among a myriad of other possibilities. 53 References 1 Starr and Taggart, Biology: The Unity and Diversity of Life, Eighth Ed., Wadsworth Publishing Company, Belmont, CA, 231 (1998). 2 Ibid., 232-233. 3 Ibid., 46-48. 4 http://www.bact.wisc.edu/MicrotextBook/BacterialStructure/Proteins.html. 5 Schafer, Lothar, In Search of Divine Reality, University of Arkansas Press, Fayetteville, 73 (1997). 6 Colón, W. and Kelly, J. W., ―Partial Denaturation of Transthyretin is Sufficient for Amyloid Fibril Formation In Vitro‖ Biochemistry 31, 8564-8660 (1992). 7 Brandon and Tooze, Introduction to Protein Structure, 269-284 (1991). 8 http://info.bio.cmu.edu/courses/03231/LecF02/Lec08/lec08.html 9 Govindarajan, Sridhar and Richard A. Goldstein, ―On the thermodynamic hypothesis of protein folding,‖ Proceedings of the National Academy of Sciences USA 95, 5545-5549 (1998). 10 http://svr.ssci.liv.ac.uk/~volk/folding/Fasteventsinproteinfolding.htm 11 http://www.rcsb.org/pdb/experimental_methods.html 12 Ladokhinin, Alexey S, ―Fluorescence Spectroscopy in Peptide and Protein Analysis,‖ Encyclopedia of Analytical Chemistry, John Wiley & Sons Ltd, 5762-5779 (2000). 13 Feynman, Richard, QED, Princeton Science Library, Princeton, NJ, 14-15 (1985). 14 http://www.grc.nasa.gov/WWW/RT2001/6000/6712ansari.html 15 Pecora, Robert, Dynamic Light Scattering, Plenum Press, New York, 11-19 (1985). 16 Ibid. 17 Ibid. 18 Brookhaven Instruments Corporation, ―Instruction Manual for BI9000AT Digital Autocorrelator.‖ 19 Berne, Bruce and Robert Pecora, Dynamic Light Scattering, Dover, Mineola, NY, 2000: 10-28. 20 Ibid. 21 Shortle, David and Michael Ackerman, ―Persistence of Native-Like Topology in a Denatured Protein in 8M Urea,‖ Science 293, 487-489 (2001). 22 Ibid. 23 Fortner, April, ―Protein Chemical Denaturation and Analysis using Dynamic Light Scattering,‖ University of Arkansas Honors Thesis, Spring 2003. 24 Stites Lab Protocol, ―Nuclease Protein Preparation in λ Expression System.‖ University of Arkansas, Rev. September 2002. 25 R. S. Stock and W. H. Ray, ―Interpretation of Photon Correlation Spectroscopy Data: A Comparison of Analysis Methods,‖ J. Polymer Science: Polymer Physics Edition 23, 1393 (1985). 26 CRC Handbook for Chemistry and Physics, 1988. 27 Gast, Klaus et al., ―Compactness of the kinetic molten globule of bovine αlactalbumin: A dynamic light scattering study,‖ Protein Science 7, 2004-2011 (1998). 54