Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Use of peptide fragments for investigating calcium-binding proteins and protein folding Tsai, Frank C. S. 1995

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

Item Metadata

Download

Media
831-ubc_1995-0421.pdf [ 3.4MB ]
Metadata
JSON: 831-1.0061741.json
JSON-LD: 831-1.0061741-ld.json
RDF/XML (Pretty): 831-1.0061741-rdf.xml
RDF/JSON: 831-1.0061741-rdf.json
Turtle: 831-1.0061741-turtle.txt
N-Triples: 831-1.0061741-rdf-ntriples.txt
Original Record: 831-1.0061741-source.json
Full Text
831-1.0061741-fulltext.txt
Citation
831-1.0061741.ris

Full Text

USE OF PEPTIDE FRAGMENTS FOR INVESTIGATING CALCIUM-BINDING PROTEINS AND PROTEIN FOLDING By FRANK C. S. TSAI B. Sc., University of Toronto, 1991 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF CHEMISTRY We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA JULY 1995 © Frank C. S. Tsai, L995 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department The University of British Columbia Vancouver, Canada DE-6 (2/88) 11 Abstract Two projects are described that use synthetic peptides to study protein structure. The EF-hand C a 2 + binding protein calbindin Dcj^ was studied by synthesizing and characterizing peptides corresponding to each C a 2 + binding domain. The peptide called calb 1 corresponds to the site I binding domain and the peptide called calb 2 corresponds to site II. Circular dichroism (CD) spectra in collaboration with AV and calorimetry experiments determined by other groups indicated that calb 1 alone does not bind C a 2 + . Calb 2 binds C a 2 + in a mechanism whereby one C a 2 + enables two peptides to be folded into a helical conformation in an apparent dimer. A 1:1 mixture of calb 1 and calb 2, called calb 1/2, binds two equivalents of C a 2 + into a helical conformation in an apparent heterodimer. The mechanism appears to be stepwise in which calb 2 binds the first C a 2 + and calb 1 binds the second C a 2 + . This indicates a positive cooperative effect conferred by calb 2 onto calb 1. Our results are consistent with previous work on the intact calbindin Dg^ indicating a strong dependence on favourable electrostatic interactions between C a 2 + and side chain ligands for high binding affinity, and positive cooperativity between the two binding domains. CD experiments of calb 2 and calb 1/2 in the presence of 0.1 M KC1 showed a smaller increase in helical content upon C a 2 + addition when compared to the absence of salt. The results in the presence of 0.1 M KC1 corroborate the importance of electrostatic interactions in the C a 2 + binding affinity of the peptides. A peptide called aac-heme, which corresponds to helices F and G (residues 80 to 108) of the a-chain of horse methemoglobin, was synthesized and characterized by C D spectroscopy. The sequence of aac-heme adopts a structure, called an a,a-corner in the intact protein, which has been hypothesized to be a protein folding initiator (Efimov, A . V . (1984) FEBS Lett. 166, 33-38). We tested Efimov's hypothesis by analysing the C D spectra of aac-heme in aqueous buffer and trifluoroethanol (TFE) solutions. Aac-heme is moderately helical in aqueous buffer and this helicity is concentration dependent. This indicates that Ill aac-heme aggregates to compensate for the lack of tertiary interactions which would otherwise be present in the intact protein. Aac-heme is fully helical in 60% TFE solution suggesting that a hydrophobic environment enhances the helicity of this peptide. These observations are consistent with a folding mechanism for the native hemoglobin whereby a hydrophobic collapse occurs first, followed by the formation of secondary structure. The results suggest that the a,a-corner could well initiate the secondary structural formation in hemoglobin and that this initiation would be even more pronounced if it was preceded by a hydrophobic collapse of the unfolded hemoglobin. Thus, the a,oc-corner may be integral in the folding pathway, as Efimov proposed. iv Table of Contents Abstract ii List of Figures vii List of Tables ix Abbreviations x Acknowledgements xii Dedication xiii CHAPTER 1: Calcium-Binding Fragments of Calbindin D 9 k 1 Introduction 1.1 Calcium-Binding Proteins 2 1.2 The "EF-Hand" Calcium-Binding Domain 3 1.3 Calbindin D9k 6 1.4 Calbindin Dc^ Fragments 11 1.5 Our Strategy and Goals 13 1.6 Our Methodology 14 2 Experimental 17 3 Results 3.1 Concentration Studies 18 3.2 C a 2 + Induced Helicity 21 3.3 C a 2 + Induced Volume Changes 27 3.4 C a 2 + Induced Enthalpy Changes 29 4 Discussion 4.1 Concentration Studies 30 4.2 C a 2 + Binding of Fragments in the Absence of Salt 30 4.3 Supporting Evidence for the Proposed Binding Mechanism 36 V 4.4 Evidence Against the Proposed Binding Mechanism 39 4.5 Summary: C a 2 + Binding in the Absence of Salt 41 4.6 C a 2 + Binding of Fragments in 0.1 M KC1 42 4.7 Evidence for Proposed Binding Mechanism 46 4.8 Summary: C a 2 + Binding in 0.1 M KC1 48 5 Conclusion 5.1 Summary of Results 5 0 5.2 Future Studies 50 References to Chapter 1 52 CHAPTER 2: The a,a-Corner Super-Secondary Structure Motif as Protein Folding Initiator 1. Introduction 1.1 The Protein Folding Problem 5 8 1.2 The a,a-Corner as Protein Folding Initiator 61 1.3 The oc,a-Corner from the a-Chain of Horse Methemoglobin 66 2. Experimental 6 9 3. Results 70 4. Discussion 4.1 Concentration Studies 75 4.2 T F E Studies 76 4.3 Protein Folding Initiation 7 7 4.4 Protein Folding Models 78 5. Conclusion 5.1 Summary of Results 80 5.2 Future Studies 80 References to Chapter vu List of Figures Figure Page 1 Schematic and backbone representation of the EF-hand of C a 2 + binding domain of 4 carp muscle parvalbumin. 2 Consensus sequence of EF-hands. 5 3 Schematic drawing of calbindin D ^ . 6 4 C a 2 + coordination octahedra of calbindin D^. 8 5 Definition of binding constants: macroscopic binding constants, K i and K2, and 9 the microscopic binding constants, K\, K\\, K\j\ and A/rj,l-6 Concentration independence of calb 1 and 2 in the absence of salt. 19 7 Concentration independence of calb 1 and 2. 20 8 C a 2 + titration of calb 1 (26 uM), calb 2 (25 uM) and calb 1/2 (26 uM). 22 9 Representative CD spectra of calb 1 (26 uJVI), calb 2 (25 UJVI) and calb 1/2 (26 | iM) 23 in excess C a 2 + . 10 C a 2 + titration of calb 1 (20 uM), calb 2 (39 fiM) and calb 1/2 (24 fiM). 25 11 Representative CD spectra of calb 1 (20 ^M) , calb 2 (39 uM) and calb 1/2 (24 uM) 26 in excess C a 2 + . 12 Volume change profile of calb 1, calb 2, calb 1/2 and intact calbindin D9k. 28 13 A schematic representation of a possible mechanism of C a 2 + binding of calb 2. 34 14 A schematic representation of the possible mechanism of C a 2 + binding of calb 1/2. 35 15 Reaction equations of C a 2 + binding of calb 1, calb 2 and calb 1/2 in the absence of 42 salt. 16 Schematic representation of a possible mechanism of C a 2 + binding of calb 1/2. 45 17 Reaction equations of C a 2 + binding of calb 1, calb 2 and calb 1/2 in 0.1 M KC1. 49 18 Criteria required for solving the protein folding reaction. 59 19 Schematic diagram of a,a-corner. 63 Vlll 20 Schematic representation of a right-handed cc,a-corner with a short connection 64 consisting of two peptide units. 21 Amino acid sequence of the a-chain of horse methemoglobin. 67 22 Schematic structure of hemoglobin indicating the location of the F and G helices. 68 23 C D spectrum of aac-heme (93 mM) in 50 mM sodium borate, 0.1 M NaCl, pH 7, 71 4° C. 24 Circular dichroism spectrum of aac-heme (12.7 uM) in 60% TFE, 4°C. 72 25 Helicity of aac-heme as a function of TFE percentage. 73 26 Helcity of aac-heme as a function of concentration. 74 ix List of Tables Table Page 1 Ellipticity of calb 1, calb 2, and calb 1/2 in 20 mM PIPES, pH 7, 4° C. 21 2 Ellipticity of calb 1, calb 2, and calb 1/2 in 20 mM PIPES, pH 7, 0.1 M KC1, 4° 24 C . 3 Volume changes of calb 1, calb 2, calb 1/2 and calbindin D 9 k in 20 m M PIPES, 27 p H 7 , 0 . 1 M K C 1 , 20° C. 4 Enthalpy change and association constant of calb 1, calb 2, calb 1/2 and intact 29 calbindin determined at 21° C. 5 Values of 6222 for calb 1, calb 2, and calb 1/2 in the absence of salt, pH 7, 20 m M 33 PIPES X Amino Acids: A Ala C Cys D Asp E Glu F Phe G Gly H His I lie K Lys L Leu M Met N Asn P Pro Q Gin R Arg S Ser T Thr V Val W Tip Y Tyr Abbreviations Alanine Cysteine Aspartic Acid Glutamic Acid Phenylalanine Glycine Histidine Isoleucine Lysine Leucine Methionine Asparagine Proline Glutamine Arginine Serine Threonine Valine Tryptophan Tyrosine Others: CD Circular dichroism xi DJPEA Diisopropylethylamine DTT Dithiothreitol H B T U 2-( lH-benzotriazol-1 -yl) -1,1,3,3-tetramethyluronium hexafluorophosphate HOBt 1-Hydroxybenzotriazole H P L C High performance liquid chromatography PIPES Piperazine-N, N'-bis(2-ethanesulfonic acid) TFA Trifluoroacetic acid TFE Trifluoroethanol Xll Acknowledgements I would like to acknowledge the tremendous understanding and guidance provided by John Sherman during my stay at U B C . This was, and still is, a great learning process of myself and the world around me, and he has contributed a great deal to that learning. To my colleagues at the Department of Chemistry, past and present, from whom I've learned a great deal also, I hope the best for all. Last, but definitely not least, to my friends and family, in Vancouver and elsewhere, who have provided unconditionnel love and support; I hope you'll always be around, I wouldn't be here otherwise. Xlll This thesis is dedicated to my parents, and to my parents' parents. xiv "Into the core of Nature"-O Philistine-"No earthly mind can enter." The maxim is fine; But have the grace To spare the dissenter, Me and my kind. We think in every place We're at the center. "Happy the mortal creature To whom she shows no more Than the outer rind," For sixty years I've heard your sort announce It makes me swear, though quietly; To myself a thousand times I say: A l l things she grants, gladly and lavishly; Nature has neither core Nor outer rind, Being all things at once, It's yourself you should scrutinize to see Whether you're center or periphery. Johann Wolfgang von Goethe 1 Chapter 1: Calcium-Binding Fragments of Calbindin D9k A day like today I realize what I've told you a hundred different times - that there is nothing wrong with the world. What's wrong is our way of looking at it. Henry Miller 2 1 Introduction 1.1 Calcium-Binding Proteins The important role of calcium ions in many physiological functions has been known since the late 19th century (Cheung, 1980). Sydney Ringer (1883a, b) showed their indispensability in embryonic development, cell adhesion, and the contractility of the heart. Further research reinforced the importance of calcium ions in muscle contraction, blood coagulation, secretion of fluids and electrolytes, cell growth, and bone and muscle development. Although calcium ions were regarded as essential components of these processes, the modulating factor that links them with these physiological functions was unknown. Then, in the middle of this century, researchers discovered calcium modulated proteins such as myosin ATPase, the troponin complex, calmodulin and parvalbumin (Heizmann, 1991). With these, scientists determined that the role of calcium-binding proteins is to mediate calcium ions in a complex messenger system for eliciting physiological processes. Calmodulin for example, regarded as the prototypical calcium-modulated protein, interacts with at least 15 different proteins (Seamon & Kretsinger, 1983). In other examples, troponin C and parvalbumin act as calcium modulated components of a switch that regulates muscle contraction. In all these cases, researchers recognized calcium-binding proteins as the only means to adequately regulate calcium ions for proper physiological functions. Cytosolic calcium-binding proteins play the mediating role in physiological functions by responding to a C a 2 + stimulus across the plasma membrane. Intracellular C a 2 + concentration is approximately l O - 7 M while the extracellular concentration is three to four orders of magnitude higher. This disparity allows the influx of C a 2 + into the cell to act as a trigger for calcium-dependent enzymes and/or other calcium dependent processes. The transient increase in intracellular C a 2 + concentration is restored to basal levels by extrusion or sequestration of C a 2 + by calcium-binding proteins. The proteins directly bind C a 2 + or begin a cascade of events with other calcium-dependent proteins to maintain C a 2 + homeostasis. A l l 3 known cytosolic calcium-binding proteins function in part or in whole to control or maintain C a 2 + concentration in this way. 1.2 The "EF-Hand" Calcium-Binding Domain The functional similarity of various cytosolic calcium-binding proteins is paralleled by the structural similarity in their calcium-binding sites (Heizmann & Hunziker, 1991). The sites consist of a helix-loop-helix super-secondary structure. The "EF-hand" binding domain was first described by Kretsinger in 1973 based on the x-ray crystal structure of carp muscle parvalbumin (Kretsinger & Nockolds, 1973). The thumb and forefinger of the right hand extended at an approximate right angle symbolizes the orientation of the E and F a-helices of parvalbumin (Figure 1). The calcium ion is coordinated octahedrally by side chains or backbone carbonyls in a loop between the helices. A l l known intracellular calcium-binding proteins have this calcium-binding domain. In 1975, Kretsinger (Kufty & Kretsinger, 1975; Heizmann, 1991) hypothesized that these C a 2 + modulated proteins, in contrast to extracytosolic calcium-binding proteins, all contain the EF-hand and belong to one homologous family, the calmodulin superfamily. Thus far, no exceptions have been found. Subsequent classification showed that the EF-hand homologous family of proteins can be divided into 31 subfamilies (Nakayama & Kretsinger, 1994). Our protein of interest, calbindin D9k, belongs to the subfamily S-100. This is the only subfamily that contains proteins with only two EF-hand calcium-binding sites, the first of which varies slightly from the consensus sequence, described below. 4 Figure 1. Schematic and backbone representation of the EF-hand C a 2 + binding domain of carp muscle parvalbumin. Reproduced from Kawasaki & Kretsinger, 1994. Kretsinger analyzed over a thousand known sequences. He developed a useful mnemonic for identifying EF-hands (see Figure 2) and generated a consensus sequence (Kawasaki & Kretsinger, 1994). The typical sequence is 29 residues long with the E helix comprising residues 1 to 10, the loop comprising residues 10 to 21, and the F helix comprising residues 19 to 29. The residues indicated by n, positions 2, 5, 6, 9, 17, 22, 25, 26 and 29, are usually hydrophobic. They form a hydrophobic core and interact with homologous residues of 5 a second EF-hand domain, related to the first by an approximate twofold rotation axis. Isoleucine, leucine or valine at position 17 contacts the loop to the hydrophobic core. The residues indicated by *, positions 3, 4, 7, 8, 11, 13, 19, 20, 23, 24, 27, and 28, are variable, often hydrophilic. Glycine at position 15 permits a sharp bend in the calcium-binding loop. Position 1 has a strong consensus for glutamic acid, but it is not invariant. The calcium ion ligands are indicated by the corresponding vertices of its octahedral coordination sphere. The vertices x, y, z and -z are the side chains of residues 10, 12, 14, and 21, respectively. Vertex -x is often a water molecule and -y is the backbone carbonyl of position 16. Of its two calcium-binding sites, calbindin Dg^ contains one consensus sequence and one variant consensus sequence. CANONICAL l o o p h e l i x E h e l i x F 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 E n * * n n * * n X * Y * Z G # I-X * *-Z n * * n n * * (n) Figure 2. Consensus sequence of EF-hands. Key: E is glutamic acid; G is glycine; I is isoleucine; n denotes hydrophobic residues; * denotes a hydrophilic residues; X, -X, Y, Z, and -Z are the coordinates of the C a 2 + ligation; and # denotes a backbone carbonyl coordination to the -Y vertex of C a 2 + ligation. Reproduced from Kawasaki and Kretsinger, 1994. 6 1.3 Calbindin D9k Calbindin D9k is the smallest member of the calmodulin superfamily, which contains 75 amino acids and weighs 9000 Da. It is found predominantly in mammalian intestine epithelial cells and is believed to aid in C a 2 + resorption or function as a C a 2 + buffer in these cells (Schachter, 1980; Levine & Williams, 1980; Christakos et al, 1989). A high resolution x-ray crystal structure (Szebenyi & Moffat, 1986) reveals that each of the two calcium-binding sites consists of about half of the protein (Figure 3). Figure 3. Schematic drawing of calbindin D9k- Reproduced from Johansson et al., 1993. The first helix-loop-helix comprises amino acid residues 3 to 36 and the second comprises residues 45 to 75 (Linse et al., 1987). A nine residue linker from position 37 to 44 connects these two sites. The sequence of the N-terminal site (site I) differs slightly from the consensus 7 EF-hand of the calmodulin superfamily. The N-terminal site contains two insertions that are absent in the consensus sequence: alanine at position 15 and asparagine at position 21 (Herzberg & James, 1985). Due to the variation in the N-terminal site, it is often called the "pseudo EF-hand." The C-terminal site (site II) is archetypical and contains 12 amino acids with identical coordination patterns to other EF-hands (Herzberg & James, 1985). The coordination sphere around the C a 2 + is roughly octahedral for both sites (See Figure 4) (Szebenyi & Moffat, 1986). The backbone carbonyls of Alal4 , Glul7, Aspl9 and Gln22 are the x, y, z and -y ligands of the first calcium-binding site, respectively. Water coordinates at -x and the side chain carboxylate of Glu27 at -z (Figure 4). In the second site, side chain carbonyls of Asp54, Asn56, Asp58, Glu60 and Glu65 are the x, y, z, -y and -z ligands, respectively. A water molecule is the -x ligand. 8 Cdcium 1 l - l Domain O J Z 7 0 / S ' ( - Z ) Cdcium 2 II—fV Domain Ou 65 Ots (-z) Figure 4. Ca 2 + coordination octahedra of calbindin D9k. Ligands are roughly octahedral, as indicated by the deviation from the ideal octahedron. Reproduced from Szebenyi & Moffat, 1986. In 1987, Linse et al. performed Ca 2 + titrations using equilibrium fluorescence and Ca 2 + selective electrodes to show that the two sites bind with K\ = 2.2 x 108 M"1 and K2 = 3.7 x 108 M- 1 (See Figure 5). 9 ^ 1 K2 P „ ^ P(Ca + 2 )! ^ P(Ca + 2 ) 2 Ki=Ki + KU K2 = KlKlul{Ki + KYL) AG\ot = -KTln(KlK2) Figure 5. Definition of binding constants: macroscopic binding constants, K\ and K2, and the microscopic binding constants, K\, Ku, K\JI and K\i\. From Linse, 1987. Furthermore, the two sites interact with each other. Their cooperativity can be defined as the difference in free energy change of binding an ion to a site in the presence (AGN>I or A G 1 ' 1 1 ) and absence (AG 1 1 or AG 1 ) of an ion in the other site (Carlstrom & Chazin, 1993; Linse et ah, 1987): 10 A A G 0 = A G 0 i , i i - A G 0 i = A G 0 n ( i - A G 0 i i eq. 1 Linse et al. (1987, 1991a, 1994) used N M R techniques and found A A G to be about -6.9 kJ/mol in low salt and -5.1 kJ/mol in physiological ionic strength (0.15 M KC1), indicating a positive cooperativity between the two sites. Calbindin D9k essentially loses cooperativity in 1 M KC1, yet retains its structure and calcium-binding function, albeit with a decrease in Ka of five orders of magnitude (Kesvatera et al., 1994). An essential element to understanding the binding process and establishing the molecular basis for cooperativity is the half-saturated protein. However, the cooperativity and the similar affinities of the two sites impede direct investigation of the this species. Researchers have circumvented this problem in various ways. Carlstrdm and Chazin (1993) used the E65Q-P43M 1 mutant protein to obtain the P(Ca 2 + ) i state. Using N M R techniques, they observed that when the first C a 2 + is bound to site I, this site adopts a conformation and dynamic state similar to the fully calcium-loaded protein state. Only minor changes occur in site II until the second C a 2 + is bound. They concluded that C a 2 + binding in site I does not dramatically affect site II. In contrast, Akke et al. (1991) used the sequential binding of cadmium ions (Vogel et al., 1985) to obtain P(Cd 2 + ) i and found that the half-saturated form is more similar to the saturated form (Kordel et al., 1989; Skelton et al., 1990a) than to the apo-form (Skelton et al, 1990b). They consider that a large entropic effect from binding of the first C d 2 + to site II contributes to the cooperativity of binding the second C d 2 + to site I. In this case, C a 2 + binding in site II does dramatically affect the structure of site I. Subsequent 1 5 N N M R relaxation studies supported these conclusions. Skelton et al. (1992b) identified that the largest dynamic changes occur in the binding loops upon calcium binding. They also observed 1 E65Q is standard notation for mutants with glutamic acid (E) at position 65 mutated to glutamine (Q). In general, X n Y denotes residue X at position n is mutated to residue Y . In E65Q-P43M, there are two mutations. 11 larger chemical shift changes upon direct binding at one site than that induced by binding in the other site, and the magnitudes of the chemical shift changes associated with apoprotein —> P(Cd 2 + ) i are larger than those associated with P(Cd 2 + ) i —» P(Cd2 +)2. The implication being the binding of the first ion causes most of the structural changes in the protein. Akke et al. (1993) observed dramatically different backbone dynamics of the two binding loops. In the apo state, the site I loop has substantially smaller fluctuations on the picosecond to nanosecond time scale than the site II loop, and the effect of ion binding on these fluctuations is substantially smaller in loop I than in loop II. They attribute the different dynamics and fluctuations to the fundamental effect of the two amino acid insertions in the pseudo EF-hand over the consensus sequence. In a creative study, Brodin et al. (1990) engineered mutants as follows: The mutants are designated "1:1", indicating both sites have the site I (pseudo EF-hand) sequence; "11:11", indicating both sites have the site II (consensus EF-hand) sequence; and "11:1", indicating that site I has the site II sequence and vice versa. They found that the total free energy of binding, AGtot> is in the order: wild type > II:II > 1:1 ~ 11:1. The free energy of cooperativity, AAG, has an identical rank order. Brodin et al. concluded that the context of the binding site is important for determining its C a 2 + binding. 1.4 Calbindin D(>k Fragments The investigations of the half-saturated calbindin Dg^ just described involved the whole protein or mutants thereof. Although informative, the presence of two cooperative binding-sites inherently perturb each other, making individual study of each site difficult. Two strategies have been traditionally adopted to overcome this: the use of protein fragments from enzymatic cleavage or chemical cleavage, and preparing the fragments synthetically using 12 peptide synthesis. Both areas have provided fertile grounds for investigating many calcium-binding proteins. The synthetic approach has been used extensively for troponin C (Shaw et al., 1991a, b, 1992a; Kay et al, 1991; Ngai & Hodges, 1992), calmodulin (Foffani et al, 1991; Reid, 1987) and phosphorylase kinase (Newsholme et al, 1992). Currently, three reports are available for studying the individual calcium-binding domains of calbindin Dcjk- Tsuji and Kaiser (1991) synthesized a series of 37-residue analogues of site I and studied the change in helicity upon C a 2 + addition. They observed helix induction for the natural sequence and several mutants that were designed to stabilize helix packing formation. Finn et al. (1992) introduced a unique methionine residue by engineering the mutant P43M . Cyanogen bromide (CNBr) treatment yielded two fragments: F l , residues 1-43 and F2, residues 44-75. The structure of the F1-F2 dimer is essentially identical to the P43M mutant and the binding constants (10 5 M" 1 ) for the individual fragments are several orders of magnitude smaller than for the intact protein. Finally, Linse et al. (1993) engineered the triple mutant L39C-P43M-I73C to form a disulfide bond between the two subdomains. Cleavage with CNBr gave two halves, cysFl and cysF2, each containing one cysteine mutation. They studied the calcium-binding and stability of the homodimer and the heterodimer, as well as F l and F2 of Finn et al. (1992). They found that the heterodimer F l + F2 forms quickly upon mixing of (Fl)2 and (F2)2 in the presence of C a 2 + . This reaction is much slower when one or both homodimers contain cysteine residues without the disulfide reducing agent DTT (dithiothreitol). The disulfide bonds between identical monomers inhibit or prevent heterodimer formation. The covalent homodimers (cysFl)2 and (cysF2)2 each bind two C a 2 + , but much less favourably than either P43M or L39C-P43M-I73C. Lastly, critical for fragment studies, Linse et al. concluded that each EF-hand will form homodimers in the presence of C a 2 + to avoid exposure of the hydrophobic surface to the aqueous milieu. They further state that "because the disposition of hydrophobic and charged residues is not optimized for 13 interaction with an identical EF-hand, this interaction is not nearly as strong as with the other EF-hand." 1.5 Our Strategy and Goals Our strategy was to use peptide fragments of calbindin D9k to study the C a 2 + binding of each EF-hand. Since it is known that the helical content of this protein increases upon C a 2 + binding (Dorrington et al., 1978; Chiba et al., 1983), we have monitored the helicity of the peptides with incremental addition of C a 2 + . In collaboration, two other research groups have determined the change in volume and enthalpy of binding of the identical peptides. References to their A V and AH results will be used in the Results and Discussion. In studying the C a 2 + binding of each EF-hand, we had set goals that reflected the observations and conclusions of Linse et al. (1993) who showed that the peptides containing one EF-hand are likely to exist in solution as homodimers in the presence of C a 2 + within concentration ranges of both circular dichroism (CD) and N M R experiments. Similarly, a mixture of the two different EF-hands are likely to exist in solution as a heterodimer. This would be consistent with the observation that most EF-hands exist in pairs (Brodin et al., 1990; Nakayama & Kretsinger, 1994). Thus, we designed the experiments with the following goals: I) To determine the relative C a 2 + affinity of the dimeric species. Based on current information, one can expect that the C a 2 + affinity should be in the order: (site I+site II) > (site II) 2 > (site 1)2- This is based on the fact that the heterodimer has a "natural" disposition of hydrophobic and charged residues compared to the homodimers, as discussed by Linse et al. (1993). The site II homodimer is predicted to have a greater C a 2 + affinity than the site I homodimer based on the electrostatics and the binding pattern of site I amino acids versus site n (Szebenyi & Moffat, 1986). The total electrostatic charge in site I is -1 compared to -6 in site 14 II (Finn et al, 1992). Also, four of six ligands on site I are backbone carbonyls compared to none in site II, where they are all amino acid side chains. Thus, we hoped to support Tsuji & Kaiser's suggestion that the pseudo EF-hand, site I, "may require a specific sequence to form a rigid conformation and arrange the ligands properly for calcium binding", unlike the consensus EF-hand of site II. 2) The effect of salt on C a 2 + affinity for the dimers. Salt screens electrostatic interactions, and has been used extensively in experiments and computer simulations (Linse et al., 1991a, 1994; Svensson et al, 1991; Kesvatera et al., 1994). The salt effects on the intact protein and mutants are known, but its effects on homodimers are less clear. In the one study with homodimers, Linse et al. (1993) found a dramatic effect on AGtot for both (cysFl)2 and (cysF2)2. It is believed that electrostatics plays an important role in the cooperativity of the two binding sites. Current information shows a larger effect on site II C a 2 + binding than site I. Thus, we expected that salt would affect the site II homodimer greater than the site I homodimer. 1.6 Our Methodology We used the fragmentation strategy to investigate the calcium binding of the two sites. The protein fragments are synthetic and constitute the two EF-hand domains of calbindin D9k. The sequence of the site I fragment, called calb 1, is: [ A c ] - S - P - E - E - L - K - G - I - F - E - K - Y - A - A - K - E - G - D - P - N - Q - L - S - K - E - E - L - K - L - L - L -Q-T-E- [NH 2 ] which are residues 2 to 35 of the intact protein. The sequence of the site II fragment, called calb 2, is: 15 [ A c ] - T - L - D - E - L - F - E - E - L - D - K - N - G - D - G - ^ ^ [NH 2 J which are residues 45 to 74 of the intact protein. We acetylated the N-termini and amidated the C-termini of both peptides to mimic the lack of charges in the intact protein and eliminate any charged-group interaction with the helix macro-dipole (Shoemaker et al., 1987; Wada, 1976). The nine residue linker of the intact protein and the N-terminal lysine and C-terminal glutamine were deleted because they were not deemed to be part of either site I or II binding domain (Szebenyi & Moffat, 1986). We studied the interaction of calb 1, calb 2 and calb 1/2 (a 1:1 mixture of calb 1 and calb 2) with C a 2 + by CD. We monitored ellipticity at 222 nm as an indication of helicity of the peptide since it is a well accepted method based on CD studies on proteins (Greenfield & Fasman, 1969; Adler et al., 1973). Previous studies on calbindin D 9 k have shown C D to be a viable method of detecting the structural changes upon C a 2 + binding (Tsuji & Kaiser, 1991; Dorrington et al., 1978; Chiba et ah, 1983). The salt effects were studied by performing experiments without salt or with 0.1 M KC1. Aggregation of the individual peptides was studied by monitoring molar ellipticity with varied peptide concentration. The theory of AV is based on measuring the decompression (or release) of electrostricted water molecules upon the coordination of solutes by the ligand(s) in solution (Kupke & Fox, 1989). In aqueous media, water dipoles are highly compressed in the strong electric fields immediately adjacent to ions. Large changes in volume occur when multiple charged ligands coordinate to a metal ion by causing a local collapse of the open structure of normal water (Kupke & Fox, 1989). Professor Donald Kupke of the Department of Biochemistry, University of Virginia, performed Admeasurements by titrating C a 2 + on calb 1 and 2 samples using a magnetic-suspension densimeter (Gillies & Kupke, 1988). 16 The measurement of AH was performed by Karen Alessi and Professor Luis Marky of the Department of Chemistry, New York University. Calorimetric titration of C a 2 + at 21°C with the same peptide samples gave values for AH and Ka. C D experiments were not performed on calbindin Dg\^ due to lack of C a 2 + free protein samples, however AV, AH and Ka measurements were done on the intact protein. 2 Experimental 17 A l l reagents were used without further purification. PIPES buffer, guanidine hydrochloride, trifluoroethanol and analytical grade CaC^ were from Aldrich. Analytical grade KC1 was from B D H . Calb 1 and calb 2 were both synthesized by an A B I 430 Peptide Synthesizer using t-Boc chemistry and cleaved by HF. We thank the lab of Ian Clark-Lewis for the synthesis and cleavage of the peptides. They were both purified by reverse phase HPLC using Phenomenex C-l8 columns (4.6 mm x 250 mm or 22.5 mm x 250 mm) with eluents A (water/0.1% TFA) and B (acetonitrile/0.05% TFA). A gradient of 65% A to 60% A over 50 minutes was used to purify calb 1. A gradient of 60% A to 50% A over 40 minutes was used to purify calb 2. The peptides were determined to be greater than 98% pure by analytical reverse phase HPLC. Electrospray mass spectrometry was used to identify calb 1: 3918.94 ± 1.7, calcd 3917.45. Liquid secondary ion mass spectrometry was used to identify calb 2: m/z 3500.89 (M+H) + , calcd 3499.89. Circular dichroism spectra were acquired on a Jasco J-710 spectropolarimeter equipped with Neslab RTE-210 circulating water bath. A l l stock solutions, buffers and C D samples were filtered through a 0.45 um filter prior to data acquisition. A l l spectra were acquired at 4° C after a 10 minute equilibration. A l l spectra were acquired as an average of three scans using a cuvette with a 1 mm path length. Concentrations of calb 1 were measured using tyrosine absorbance at 275 nm in 6 M guanidine hydrochloride (Brandts & Kaplan, 1973). Concentrations of calb 2 were measured by amino acid analysis using a norleucine internal standard. Other conditions are specified in the individual figures. Experimental details of AV and AH measurements are available from Dr. Kupke and Dr. Marky, respectively. 18 3 Results 3.1 Concentration Studies The simple interpretation of peptide secondary structure requires that they have a definable aggregation state in solution. We monitored the helicity at 222 nm with varied peptide concentrations. An increase in helicity with increasing concentration would indicate self-aggregation. Figures 6 and 7 show the independence of helicity of calb 1 and calb 2 with increasing concentration under no salt and 0.1 M KC1 conditions. For both peptides and under both conditions, there is no significant change in helicity. This indicates that, within the concentration range of 10 to 250 f iM, at 4 °C, the aggregation state of calb 1 and calb 2 does not change. We assume that the aggregation states are constant, monomeric in the absence of C a 2 + . Dimers of calb 2 form in the presence of C a 2 + (discussed later). Dimerization of EF-hands in the presence of C a 2 + is supported by literature precedents (Linse et al, 1993) and the propensity of C a 2 + loaded EF-hands to exist in pairs (Brodin et al, 1990; Nakayama & Kretsinger, 1994). We performed all subsequent experiments within the concentration range of 10 to 30 u M . Concentration studies of calb 1/2 were not performed. The aggregation state for calb 1/2 will be discussed in detail. 19 15000 o E £ e o o> CO T J , CM <M CM & © 10000 -\ 5000 I I I I I I I I I | I I I I 50 100 150 Concentration (uM) I ' I I ' I " 200 250 Figure 6. Concentration independence of calb 1 and calb 2 in the absence of salt. Conditions are pH 7, 20 m M PIPES and 4° C. 20 15000 o E E o O) <D 2 . CM CM CM et "5 10000 5000 H 4 0 0 Concentration, (uM) Figure 7. Concentration independence of calb 1 and calb 2. Conditions are pH 7, 20 m M PIPES, 0.1 M K C 1 and 4° C. 21 3.2 C a 2 + Induced Helicity Figure 8 shows the change in helicity of the calbindin D 9 k fragments upon addition of C a 2 + . Table 1 shows the change in helicity for the three species in the absence of C a 2 + and in the presence of C a 2 + . Representative spectra are shown in Figure 9. Calb 1 has low helicity in both the absence and presence of C a 2 + . Calb 2 has low helicity in the absence of C a 2 + , but high helicity when excess C a 2 + is added. Calb 1/2 has moderate helicity in the absence of C a 2 + and the helicity increases as C a 2 + is added. Table 1: Ellipticity of calb 1, calb 2, and calb 1/2 in 20 m M PIPES, pH 7, 4° C monitored at 222 nm. 9 values are in deg cm2/dmol. 0 (no Ca 2 + ) 9 (excess Ca 2 + ) AO Calb 1 -5400 -3600 1800 Calb 2 -5400 -14800 -9400 Calb 1/2 -9800 -14800 -5000 22 Figure 8. C a 2 + titration of calb 1 (26 | i M ) , calb 2 (25 u,M) and calb 1/2 (26 ( i M ) . Conditions are pH 7, 20 m M PIPES and 4° C. 23 Figure 9. Representative C D spectra of calb 1 (26 | iM) , calb 2 (25 p:M) and calb 1/2 (26 uM) in excess C a 2 + . Conditions are pH 7, 20 m M PIPES and 4° C. 24 Figure 10 shows the change in helicity for the same experiment, but in the presence of 0.1 M r KC1. Table 2 shows the change in helicity for the three peptides in the presence of 0.1 M KC1, in the absence and presence of C a 2 + . Representative spectra are shown in Figure 11. Calb 1 has low helicity in both the absence and presence of C a 2 + . Calb 2 and calb 1/2 have low helicity in the absence of C a 2 + and moderate helicity with excess C a 2 + . Table 2: Ellipticity of calb 1, calb 2, and calb 1/2 in 20 m M PIPES, pH 7, 0.1 M KC1, 4° C monitored at 222 nm. 9 values are in deg cm2/dmol. 9 (no Ca 2 + ) 9 (excess Ca 2 + ) AO Calb 1 -2800 -3000 -200 Calb 2 -3600 -8600 -5000 Calb 1/2 -3600 -5400 -1800 25 15000 o £ CM E o CD CD 2 , CM CM CM CO * CD JO 10000 H 5000 1 1 I 1 I 1 I 2 4 6 8 , I I I I I 10 12 14 16 18 20 Equivalents of Ca+2 Figure 10. C a 2 + titration of calb 1 (20 uM), calb 2 (39 UM) and calb 1/2 (24 | i M ) . Conditions are pH 7, 20 m M PIPES, 0.1 M KC1 and 4° C. Figure 11. Representative CD spectra of calb 1 (20 uM), calb 2 (39 uM) and calb 1/2 (24 uM) in excess C a 2 + . Conditions are pH 7, 20 m M PIPES, 0.1 M KC1 and 4° C. 27 3.3 C a 2 + Induced Volume Changes Figure 12 shows the volume change profile as determined by Professor Kupke. Calb 1 shows no volume change. This is consistent with our CD results that show no change in helicity in the presence of excess C a 2 + . Calb 2 shows increasing volume change to 30 mL/mol when one equivalent of C a 2 + is added. No change in AV is observed with further addition of C a 2 + . Calb 1/2 shows a steep increase in volume change to 60 mL/mol when two equivalents of C a 2 + are added. No further changes in A V were observed with further addition of C a 2 + . The intact protein shows a similar A V profile as calb 1/2: volume change increases to 60 mL/mol when two equivalents of C a 2 + are added. The data is summarized in Table 3. Table 3: Volume changes of calb 1, calb 2, and calb 1/2 and calbindin D 9 k in 20 m M PIPES, pH 7, 0.1 M KC1, 20° C. AV values given in mL/mol. 1 eqv. C a 2 + 2 eqv. C a 2 + 3 eqv. C a 2 + Calb 1 0 0 _ Calb 2 30 30 30 Calb 1/2 45 60 60 Calbindin D 9 k 30 60 60 28 -10 0 1 2 3 Equivalents of Ca+2 • Intact Calbindin D9k • Calb 2 I Calb 1 Z Calb 1/2 Figure 12. Volume change profile of calb 1, calb 2, calb 1/2 and intact calbindin D9k-Conditions are pH 7, 20 mM PIPES, 0.1 M KC1 and 20° C. From D. Kupke. 29 3.4 C a 2 + Induced Enthalpy Changes Table 4 shows the results of calorimetry studies performed by Alessi and Marky. Table 4: Enthalpy change and association constant of calb 1, calb 2, calb 1/2 and intact calbindin D 9 k determined at 21° C upon Ca 2 +addition. AH (kcal /moi) *a(M-l) Calb 1 0 0 Calb 2 -0.6 10 4 Calb 1/2 -1.9, -0.8* 10 7, 10 5 Calbindin D 9 k -1.5 6 x 10 6 *Alessi and Marky observed an apparent break after one equivalent C a 2 + in calb 1/2 and have divided the observed values for the first and second equivalents of C a 2 + . The enthalpy results are reasonably consistent with both the CD and AV results. Calb 1, when alone, creates no change in enthalpy upon C a 2 + addition, suggesting no binding of C a 2 + . The AH of calb 2 is approximately half of calb 1/2 and the intact calbindin D 9 k . The Ka of calb 2 is two orders of magnitude less than that for calb 1/2 and the intact calbindin D 9 k. Calb 1/2 and calbindin D 9 k have comparable C a 2 + binding association constants. 30 4 Discussion 4.1 Concentration Studies Previous studies have shown that, in the presence of C a 2 + , isolated EF-hands are presumably not stable, as homodimerization takes place. This has been found for peptide fragments corresponding to site III of chicken skeletal troponin C (Shaw et al., 1990, 1991a, 1992a, b), site IV of rabbit skeletal troponin C (Kay et al, 1991), and parvalbumin (Linse & Forsen, 1995; Permyakov et al., 1991). These results, combined with the dimerization observed by Finn et al. (1992) and Linse et al. (1993) for calbindin D 9 k fragments at similar concentrations as our experiments, provide reasonable evidence that our peptides also exist as dimers when C a 2 + is bound. The prevalence of EF-hand pairs lends further support for this conclusion (in 31 subfamilies of EF-hand proteins containing 996 EF-hand domains, only two domains are not paired; Nakayama & Kretsinger, 1994). The CD experiments indicate that the aggregation state does not change in the concentration range studied. Our results show that calb 1 is monomeric both in the absence and presence of C a 2 + . Calb 2 is monomeric in the absence of C a 2 + , but forms a homodimer when C a 2 + is added (discussed below). 4.2 C a 2 + Binding of Fragments in the Absence of Salt Our first goal was to determine the C a 2 + affinity of the calb 1, calb 2 and calb 1/2 in the absence of salt. C D experiments show that calb 1 has no significant change in helicity upon addition of C a 2 + . It appears not to bind C a 2 + at all. Calb 2 and calb 1/2 both bind C a 2 + as indicated by the increase in helicity in CD experiments. Results with AV, AH and Ka are consistent with the C D results. A possible explanation is that calb 1 cannot bind C a 2 + on its own, but it is activated by calb 2. This is consistent with a binding mechanism of positive 31 cooperativity. Thus, any requirements necessary for calb 1 to bind C a 2 + is apparently fulfilled by calb 2, but not by another molecule of calb 1. In contrast, calb 2 can bind C a 2 + , whether it is in the presence of calb 1 or not. Although stoichiometry was not determined by CD, AV results are consistent with the calb 2 homodimer binding one C a 2 + (AV = 30 mL/mol), since calb 1/2 and the intact protein bind two C a 2 + (each has A V = 60 mL/mol). A possible mechanism that is consistent with our results would be for monomeric calb 2 to bind C a 2 + , attain a helical conformation, which then induces association with another molecule of calb 2 without further C a 2 + binding (see Figure 13). The association of the second calb 2 may occur in concert with C a 2 + binding or stepwise. In either case, association would explain how calb 2 is essentially structureless in the absence of C a 2 + , but attains helicity that is comparable to calb 1/2 in excess C a 2 + . According to this mechanism, calb 2 binds one C a 2 + and induces dimerization with another molecule of calb 2 that does not further bind C a 2 + . This could be an example of negative cooperativity, since the second molecule of calb 2 is affected in such a way that it does not bind C a 2 + . The final structure in Figure 13 shows that the second calb 2 peptide does not coordinate Ca2+. We indicate this based on the fact that the amino acid sequence of site II has the necessary elements for Ca2+ coordination on its own. However, our data does not preclude another possible conformation shown below: Y Further analysis is required to confirm the mode of binding. 32 A central assumption in the treatment of the data for calb 2 is that the change in 6222 is indicative of both C a 2 + binding and the association of the second calb 2 to Ca 2 + , calb 2, and not only C a 2 + binding alone. Support for this comes from both our CD studies and literature precedents. For calb 1/2 at a concentration of 26 u M in the absence of C a 2 + , 6222 = -9800 deg cm 2/dmol (see Table 5). At the same concentrations, calb 1 and calb 2 have 6222 = -5400 deg cm 2/dmol. This indicates that the greater helicity of calb 1/2 is due to association alone, thereby making association to be observable by monitoring 6222 (discussed below). The same can be reasonably assumed for calb 2. Moreover, it is common practice in this field to observe the association of helical peptides by monitoring 9 2 2 2 (Osterhout et al., 1993; Hughson et al., 1991;Scholtz etal, 1991). Incorporation of this assumption implies the following reaction equation: C a+2 _ _ calb 2 calb 2 Ca + 2-calb 2 r Ca+ 2»(calb 2) 2 where both equilibria are observable by monitoring 6222- Failure of this assumption would result in an equilibrium of only the C a 2 + binding reaction observable by 0222- However, considering the results of calb 1/2, other amphiphilic helices, and the ability to account for the data, it is reasonable to accept this assumption to be valid in our system. Our results for calb 1/2 indicate that it binds similarly to the intact calbindin D9 k . The following binding mechanism could possibly explain our results. A moderately helical calb 1/2 heterodimer is already formed in the absence of C a 2 + . This explains the helical content we observed by C D in the absence of C a 2 + . It is unlikely that this moderate helicity is the result of two helical monomeric species. This is supported by comparing the concentrations of the experiments of calb 1/2 with those of calb 1 and calb 2 (Table 5). 33 Table 5: Values of 6222 for c a i D 1, calb 2, and calb 1/2 in the absence of salt, pH 7, 20 m M PIPES. 0222 values are deg cm2/dmol. Concentration (jxM) ©222 Calb 1 26 -5400 Calb 2 25 -5400 Calb 1/2 26 -9800 The fact that ©222 of calb 1/2 is about twice that of calb 1 or calb 2 at the same total peptide concentration supports the idea of increased helicity caused by association. This is further supported by Linse's conclusion (1993) that the disposition of hydrophobic and charged residues is optimized for interactions within a pair of different EF-hands (ie. calb 1/2), and that this interaction is not nearly as strong as with identical EF-hands (ie. calb 1 or calb 2). As C a 2 + is added, calb 2 binds such that the helical content of the entire heterodimer increases. When the calb 2 site is bound with C a 2 + , a cooperative interaction occurs which activates the calb 1 site to bind C a 2 + . This is consistent with AV results where calb 1/2 has the same AVas calbindin D9k, which we know binds two equivalents of C a 2 + . When excess C a 2 + is added, both sites are occupied and no further changes takes place. The CD, AV, AH and Ka results are consistent with this stepwise mechanism (see Figure 14). 34 Figure 13. A schematic representation of a possible mechanism of C a 2 + binding of calb 2. Binding of one equivalent of C a 2 + induces helicity in one equivalent of calb 2. A second molecule of calb 2 associates and subsequently folds into a structured but C a 2 + free EF-hand. The final species, drawn schematically, is asymmetrical to indicate that a second C a 2 + is not bound. Square brackets around the intermediate species indicate that it may be formed stepwise or in concert with C a 2 + binding. Cylinders represent a-helices and the solid dots represent C a 2 + . The contact area of the packing of the helices is presumed to be the hydrophobic face of each calb 2 molecule (Szebenyi & Moffat, 1986; Finn et al, 1992). 35 calb 2 L 1 > Ca +2 calb 1 Ca +2 *1 Figure 14. A schematic representation of the possible mechanism of C a 2 + binding of calb 1/2. A moderately helical heterodimer is pre-formed (dashed cylinders). Calb 2 binds first, causing an increase in helicity. Positive cooperativity induces calb 1 to bind a second equivalent of C a 2 + , thereby forming the highly helical, C a 2 + loaded heterodimer. Square brackets around the intermediate species indicate that it may be formed stepwise or in concert with C a 2 + binding. The C a 2 + binding constants were not determined with the CD data because the evidence for the binding mechanisms is presently indirect. The calorimetry data determined by Marky and Alessi show that Ka for calb 2 is less than that for the intact protein by two orders of magnitude. The Ka of calb 1/2 is comparable to calbindin D9k. These results indicate that the interactions in calb 2 are different from those in calbindin D9k, as expected. The interactions in calb 1/2 are similar to that in calbindin D9k, notably, cooperativity between site I and site II (calb 1 and calb 2) and the affinity of C a 2 + binding. This is consistent with the proposed mechanism in Figure 14. 36 4.3 Supporting Evidence for the Proposed Binding Mechanism Shaw et al., (1991a) studied the synthetic peptide fragment of site III troponin-C, SCIII. In this study, apo-SCIII shows a random coil structure. It binds in a stoichiometry of 1:2 Ca 2 + :SCIII into a fully a-helical form, indicating that the binding of a single C a 2 + enables the folding of two polypeptide chains. The authors proposed a mechanism of SCIII C a 2 + binding that starts with initial random coil structure of apo-SCIII, followed by C a 2 + binding which induces association with another molecule of SCIII. However, Shaw et al., notes that some C a 2 + binding proteins are known to undergo very little structural change upon C a 2 + binding, and thus C a 2 + binding can appear "invisible" by some spectroscopic methods. This is true for rat parvalbumin (Williams et al., 1986) and the N-terminal domain of troponin-C (Leavis et al., 1978). They verified the stoichiometry of 1:2 Ca 2 + :SCIII by N M R . For calb 2, A V results indicate that a second C a 2 + is not bound. At saturation, calb 2 has AV = 30 mL/mol compared to calb 1/2 or calbindin D9k, which has AV = 60 mL/mol. Furthermore, with excess C a 2 + , calb 2 has the same amount of helical content as calb 1/2, which binds two C a 2 + . It is likely that both units of calb 2 are helical in the homodimer. Thus, there is substantial evidence in our results to suggest a similar mechanism of C a 2 + binding for calb 2 as SCIII and the mechanism of Figure 13. Szebenyi and Moffat's (1986) crystal structure revealed a difference in rigidity of the two EF-hands of calbindin D9k, site II being more flexible than site I. The first EF-hand, which corresponds to our calb 1, provides four of the ligands to the octahedrally coordinated C a 2 + from backbone carbonyls. This is in contrast to the second EF-hand, corresponding to our calb 2, in which no backbone carbonyls provide C a 2 + ligands; in fact, five of the C a 2 + ligands are mobile side chains of amino acids. They also found that helix IV (in site U) is more flexible than helix II (in site I) due to its irregularity and the consequent lack of some of the hydrogen bonds that would be present in a regular a-helix. These two factors combine to cause greater flexibility in the second EF-hand compared to the first. The flexibility is 37 manifested by the magnitude of the temperature factors B and the mean square displacements of the backbone and side chains in the crystal structure of calbindin D9k, which are lower in the first EF-hand than in the second. Based on this analysis, one might expect calb 1 (site I) to be more preorganized for C a 2 + binding than calb 2 (site II). However, rigidity in the C a 2 + loaded protein does not necessarily preclude or require preorganization of either site when they are in the empty form. A C a 2 + bound site can be rigid but not preorganized for binding when it is empty. In fact, dynamic N M R experiments show that the apo-site II is more flexible than apo-site I, yet site II binds C a 2 + first (Akke et al, 1993; Skelton et al, 1992a). This apparent paradox can be explained by considering other factors. The electrostatic charge of site II is much greater than site I (-6 vs -1). The flexibility of site II can be the result of electrostatic repulsion between the negatively charged side-chains of Asp54, Asp58, Glu60, Glu64, and Glu65 (Skelton et al, 1992a; Linse et al, 1991a). The favourable interaction between a doubly charged cation and the anionic side chains results in a high affinity for site II and promotes the formation of secondary structure at this site. The importance of electrostatics can also be inferred experimentally, where the E65Q-P43M mutant, merely by neutralization of one negative charge in site II, causes site I to bind first (Carlstrom & Chazin, 1993), and from theoretical studies, which shows that electrostatics confer large binding energies to the protein (Wesolowski et al, 1990; Ahlstrom et al, 1989). Thus, it appears that the favourable electrostatic interactions of site II (calb 2) with C a 2 + compensate for the entropic costs that are required to conform flexible backbone and mobile side chain atoms into C a 2 + binding, as well as desolvation of C a 2 + and the ligands. On the other hand, the lower entropic costs that are apparently required by site I (calb 1) are nullified by the lack of favourable electrostatic interactions at this site. This would explain why site II binds first in calbindin D9k, and hence, the inability of calb 1 alone to bind C a 2 + while calb 2 can. What happens once the first C a 2 + is bound to site II? In their study of the P ( C d 2 + ) i species of intact calbindin D9k, Akke et al (1991) found that the first C a 2 + binds to site II causing conformational changes not only in site n, but the entire protein as well. Amide proton 38 exchange rates afford similar conclusions (Skelton et al., 1992a, 1992b). The majority of chemical shifts of the P(Cd 2 + ) i form ( C d 2 + in site II) are more similar to the P(Ca 2 +)2 form than the apo form (Akke et al., 1991). Thus, the major portion of changes in calbindin D9k upon binding of C a 2 + occurs when the first ion binds to site II. Furthermore, molecular dynamics simulation has indicated substantial electrostatic energy contribution on site I by site II (Ahlstrom et al., 1989). This implies that the C a 2 + affinity of site I is dependent on the ion binding into site II, and that the interaction upon site I of a C a 2 + bound to site II is necessary, for site I to bind C a 2 + with an affinity nearly as high as site II. In other words, site I will bind C a 2 + only after site II has bound C a 2 + ; or analogously, calb 1 binds C a 2 + only in the presence of C a 2 + bound calb 2. This situation is obviously absent in calb 1, and is therefore consistent with calb 1 not binding C a 2 + as we observed. This also explains why calb 1/2 binds C a 2 + nearly as well as the intact protein. The experiments of Akke et al. (1991) just described corroborate our results. The work of Brodin et al. (1990) (see section 1.3) on intact calbindin closely parallels our work. Their mutants may be thought of as covalent analogues of our peptides. They engineered mutants by exchanging the two calcium-binding sites within the protein and made the amino acid sequence of the two calcium-binding sites identical. Our calb 1 would roughly correspond to their 1:1 mutant and calb 2 corresponds to their II:II mutant. Brodin et al. observed that the 1:1 mutant does not bind C d 2 + sufficiently to give any N M R signals from protein-bound C d 2 + ; and it binds C a 2 + very weakly (with AGtot of 23 kJ/mol less than wild type). Likewise, our calb 1 does not bind C a 2 + . According to 1 1 3 C d and 4 3 C a N M R , their 11:11 mutant binds both C d 2 + and C a 2 + typical of EF-hands, albeit with 13 kJ/mol less free energy than the wild type. Likewise, our calb 2 binds C a 2 + . It is possible that similar factors may be responsible for both these observations. Brodin et al. concluded that the context of the binding loop strongly influences C a 2 + binding and N M R signals indicate that the mutants may have a different fold than the wild type. From our results, we may also conclude that calb 1 requires the context of C a 2 + bound calb 2 (Ca 2 +»calb 2) in order to bind 39 C a 2 + . It is possible that isolated calb 1 adopts a different folding as compared to when Ca 2 + »calb 2 is present, although this is not yet known. Likewise, as found in the U:II mutant, the context of site II appears to be less important and folding is not markedly different from the wild type. This is consistent with our results on calb 2. Unfortunately, our current data is insufficient for further analysis and comparison. N M R experiments could determine if calb 1 and calb 2 adopt different folded structure than the intact protein. 4.4 Evidence Against the Proposed Binding Mechanism Interestingly, three fragment studies reported thus far have found different results from our investigation. Tsuji and Kaiser (1991) synthesized a series of 37 residue analogues of the pseudo EF-hand of calbindin Dg^. Their results with the natural sequence for site I (called CP-1) shows an increase of 8300 deg cm2/dmol in helicity when saturated with C a 2 + with = 0.35 mM. In contrast, our calb 1 has a modest decrease of 1800 deg cm 2/dmol in helicity and essentially doesn't bind C a 2 + at all (AH = 0 kcal/mol, Ka = 0 M " 1 , Table 4). The only obvious difference is that CP-1 is three residues longer ( K l , F36 and P37) than calb 1. N M R studies on the C a 2 + saturated calbindin Dg^ (Kordel et al., 1993) have shown that helix II of the first EF-hand extends to F36 providing two additional, fully populated {i, i+4} hydrogen bonds. Furthermore, F36 has hydrophobic interaction with L6, 19 and L31 (Szebenyi & Moffat, 1986). Together, F36 and P37 form part of a 6 residue linker loop that has been found to adopt an irregular helix in crystal (Szebenyi & Moffat, 1986) and a 3io helix in solution (Kordel et al., 1993). Carlstrom and Chazin (1993) observed a very slow amide proton exchange rate (keX = 9.1 x 10' 6 s_ 1) for F36 in the C a 2 + saturated P43G mutant suggesting that it is buried in the hydrophobic core. Spectral data for K l are less clear (see Akke et al., 1993 and Kordel et al., 1990), although chemical shifts do not change significantly upon C a 2 + binding (Kordel et al., 1989; Drakenberg et al., 1989; Skelton et al., 1990b; Carlstrom & 40 Chazin, 1993). The roles of F36 and P37 (and perhaps K l ) appear to be significant. The discrepancy between Tsuji and Kaiser's and our results indicate that these residues somehow affect C a 2 + binding. It is apparently significant enough that calb 1 does not bind at all, while CP-1 binds to a similar extent as our calb 1/2. The specifc mechanisms of this effect require further investigation. Two other fragment studies show results differing from ours. Both fragments from the CNBr cleavage of the P43M mutant bind C a 2 + (Finn et al, 1992). Similarly, (cysFl)2 and (cysF2)2 bind C a 2 + to the same extent (Linse et al, 1993). In these cases, the fragments are different from ours in terms of length and sequence. F l is the natural sequence from residue 1 to 43 and F2 corresponds to residue 44 to 75. (CysFl)2 is the same as F l except with L39C mutation. (CysF2)2 is the same as F2 except with I73C mutation. As mentioned earlier, the addition of merely three residues for CP-1 appears to be important. There are nine extra residues in both F l and (cysFl)2 compared to calb 1. F l and (cysF 1)2 includes the linker region between the two EF-hands. Although N H exchange rates are fast compared to the helical and Ca 2 + binding loop regions (Skelton et al, 1992a), there are three hydrogen bonds (Kordel et al, 1993) that may contribute to the structure and binding in F l and (cysFl)2 but not calb 1. These hydrogen bonds are all amide protons to backbone carbonyls. From donor to acceptor, they are: F36 to L32, L39 to F36, and K41 to S38. Hydrophobic contacts, which include L39 to L32 and F36, and L40 to L32 (Szebenyi & Moffat, 1986), would also be present for F l and (cysFl)2but not calb 1. The combined comparison between calb 1 and CP-1, F l , and (cysFl)2 indicates a substantial contribution to the C a 2 + affinity from the linker region or a portion thereof. It is also possible that the residues excluded in calb 1 affect its intermolecular interactions (if any) adversely or create a subtle difference in folding so that it does not bind C a 2 + . It has also been suggested that the linker may be a way that positive cooperativity is transmitted from site II to site I (Skelton et al, 1992a). A short P-sheet interaction between the two binding loops may also be responsible (Skelton et al, 1992a; Carlstrom & Chazin, 1993). Current data is insufficient to further assess these possibilities. 41 The fragment studies just described are also inconsistent with the results of the investigations of the intact proteins described in section 4.3. The subtle differences in the constituents of the fragments apparently cause large differences in C a 2 + binding behaviour. It appears that the intricacies of the C a 2 + binding of calbindin D9k are by no means easily discernable. 4.5 Summary: C a 2 + Binding in the Absence of Salt Our results with calb 1, calb 2 and calb 1/2 are consistent with the following conclusions: 1) Calb 1 requires the presence of Ca 2 +»calb 2 in order to bind C a 2 + . Calb 1 cannot bind C a 2 + on its own. This may be described as a contextual factor (Brodin et al, 1990) or positive cooperativity induced by Ca 2 + , calb 2 (Linse et al., 1987). 2) Calb 2 binds C a 2 + independently of calb 1. One equivalent of C a 2 + is sufficient to enable two calb 2 molecules to fold into high helical content. The second molecule of calb 2 does not bind C a 2 + . 3) Favourable electrostatic interactions contribute significantly to the better C a 2 + binding ability of calb 2 than calb 1. The electrostatic effects are significant enough to compensate for or overcome the large entropic costs associated with C a 2 + binding in calb 2 (or site II). 4) Minor differences in the sequence of calbindin D9k fragments cause large differences in the C a 2 + binding behaviour. Residues in the linker appear to contribute significantly to the C a 2 + affinity of site I fragments. The reaction equations for the binding mechanisms are summarized in Figure 15 below. 42 Ca +2 calb 1 _ no C a + 2 binding Ca +2 calb 2 calb 2 ±=r [ C a + 2 » c a l b 2 ] ^ - Ca + 2.(calb 2) 2 Ca +2 Ca +2 calb l«calb 2 calb l«(Ca + 2«calb 2)1 ^ ^ (Ca + 2-calb l>(Ca + 2 «calb 2) Figure 15. Reaction equations of C a 2 + binding of calb 1, calb 2 and calb 1/2 in the absence of salt. Square brackets around the intermediate species indicate that it may be formed stepwise or in concert with C a 2 + binding. 43 4.6 C a 2 + Binding of Fragments in 0.1 M KC1 Our second goal was to determine the C a 2 + affinity of calb 1, calb 2 and calb 1/2 in 0.1 M KC1. C D experiments (Figure 10) indicate that all three species have comparable helical content in the absence of C a 2 + , and overall, are less helical than in the absence of salt. Calb 1 has no change in helicity as C a 2 + is added. AV and AH are consistent with this result. The inability of calb 1 to bind C a 2 + is expected based on the results in the absence of salt. A similar binding mechanism described in section 4.2 for calb 1 is consistent with the results in 0.1 M KC1. Calb 1 cannot bind C a 2 + on its own, but it is activated by calb 2. Results of calb 1/2 indicated that a similar type of positive cooperativity is likely involved in activating calb 1 to bind C a 2 + . The presence of Ca 2 +»calb 2 is required in order for calb 1 to bind C a 2 + and the presence of another molecule of calb 1 is insufficient for fulfilling this role. Once again, calb 2 can bind C a 2 + , whether it is in the the presence of calb 1 or not. The helical content is much less in 0.1 M KC1 than in the absence of salt, which indicates a strong electrostatic effect caused by the salt. It is likely that a non-specific binding of K + causes screening of the negatively charged ligands, thereby decreasing the favourable electrostatic interactions with C a 2 + and reducing its affinity. This would explain the decrease in helical content in both the absence and presence of C a 2 + when compared with salt-free conditions. Salt may also cause screening of intramolecular hydrogen bonds or decrease the repulsion between negatively charged side chains. In these cases, salt may affect either C a 2 + binding or dimerization. In any event, our results indicate that the electrostatic interactions between the side chain ligands and C a 2 + appear to be critical, although its effects on dimerization cannot be excluded. The electrostatic effects described here do not necessarily preclude the binding mechanism represented in Figure 13, and we may consider that the same mechanism is applicable, but to a lesser degree due to salt effects. Calb 1/2 binds C a 2 + and increases helical content, although to a lesser extent than in the absence of salt. CD results indicate that a different binding mechanism is operating for calb 1/2 44 in the presence of salt than in the absence of salt. In contrast to the absence of salt where the calb 1/2 heterodimer forms before the addition of C a 2 + , salt effects have apparently prevented the association of calb 1 and calb 2. This is indicated by the helical content of calb 1/2 in the absence of C a 2 + which is comparable to calb 1 and calb 2. Salts may have disrupted the formation of secondary structure. This would prevent the formation of a hydrophobic face, which is known to be the driving force in EF-hand dimer formation, and thereby cause a decrease in helical content compared to that observed in the absence of salt. The formation of the heterodimer may occur in concert with or after C a 2 + binding, causing a nominal increase in helical content, although less than that observed for calb 2. Three explanations are possible for the observed salt effects (Figure 16): 1) the cooperative effect, AAG, has decreased, causing a reduction in the affinity of calb 1, AGi, and hence the smaller increase in helical content (case 1); 2) the C a 2 + affinity of calb 2, AGn, is decreased, which would also subsequently decrease cooperativity and affinity of calb 1 (case 2); and 3) the C a 2 + affinity of both calb 1 and calb 2 are decreased without any loss of cooperativity (the nominal increase in helicity does not necessarily preclude the maintenance of cooperativity) (case 3). A combination of all three effects is possible, since the data does not allow one to specifically discriminate between these mechanisms. Figure 16 shows a binding mechanism that is consistent with each of these possible explanations for the observed nominal increase in helicity. 45 cooperativity, AAG Calb 1 affinity, AGT Calb 2 affinity, AGjj Cooperativity, A A G Case 1 decreases same decreases Case 2 decreases decreases decreases Case 3 decreases decreases same Figure 16. Schematic representation of a possible mechanism of C a 2 + binding of calb 1/2 in the presence of 0.1 M KC1. Unlike the case in the absence of salt, calb 1 and calb 2 are not initially associated nor moderately helical. Binding of the second equivalent of C a 2 + to calb 1, facilitated by positive cooperativity from Ca 2 + , calb 2, induces helicity and association, thereby forming the final, moderately helical (Ca 2 +«calb l)»(Ca 2 +»calb 2) species. The three possible explanations for the nominal increase in helicity upon C a 2 + addition is represented in the different cases (see text). The data does not allow one to specifically discriminate between these mechanisms. Square brackets around the intermediate species indicate that it may be formed stepwise or in concert with C a 2 + binding. 46 4.7 Evidence for Proposed Binding Mechanism Our results do not definitively support one mechanism of salt effects over another. At present, we can only state that salt has dramatic effects on the helicity and C a 2 + binding of calb 1 and calb 2. Previous work also does not pinpoint specific interaction(s) of salt on intact calbindin D9k. As a consequence, we cannot specify the mechanism of salt effects, and our results can be supported by some precedents and not by others. The inability of calb 1 alone to bind C a 2 + is expected considering the results in the absence of salt. Although helical content increases nominally, CD and A V results indicate that calb 1 binds C a 2 + only in the presence of Ca 2 +»calb 2. Cooperativity, therefore, apparently remains in 0.1 M KC1, but the effect is decreased dramatically. This is consistent-with the results of Kesvatera et al. (1994) with the intact protein. High salt concentrations are observed to substantially reduce the cooperativity of C a 2 + binding in calbindin D9k (Kesvatera et al., 1994). At 1.0 M KC1, -AAG is reduced to 2 kJ/mol compared to roughly 8 kJ/mol at 2 m M KC1. The binding constants are also reduced by a factor of 5.4 pK units (-log K\K2). Similar effects were also observed for calmodulin (Linse et al., 1991b). Thus, the effects of salt are undoubtedly significant, as we observed with calb 1 and calb 2, and elsewhere by others, but the exact mechanism is still unclear. Previous work on the effect of salt on calbindin D9k has not isolated the specific interaction(s) that salt has on the C a 2 + binding properties. At present, a dielectric continuum model using a macroscopic dielectric constant of water throughout the protein agrees well between experiments and Monte Carlo simulations (Kesvatera et al., 1994; Svensson et al., 1990). Such an interior could be dramatically affected by salt, as we and others (Linse et ah, 1991; Kestavera et al, 1994; Svensson et al., 1991) have observed. However, the detailed implications of this model on C a 2 + affinity, kinetics, cooperativity and stability are still elusive. 47 Since salt effects have not been determined in detail, we may consider previous studies on electrostatic interactions on the protein and speculate on the effect of salt. Ahlstrom et al. (1989) have shown by molecular dynamics that the main source of electrostatic interaction energy in C a 2 + binding arises from the binding loops as opposed to the a-helices. In particular, site I has an equal contribution from loop II as from loop I itself, with little contribution in the reverse case (ie. site II from loop I). Since salt can screen electrostatic interactions, it is reasonable to speculate from the analysis of Ahlstrom et al. that the effect of salt would be greater for site II with a consequent effect on site I, than site I alone, and that the effect is absent for a molecule that does not bind C a 2 + . In this case, the value of A0 at 222 nm, defined as 0 s alt - 0no salt in conditions of excess C a 2 + , is larger for calb 1/2 than calb 2: 9400 vs 6200 deg cm 2/dmol. This is consistent with a salt effect that dramatically decreases C a 2 + affinity in calb 2 and hence calb 1 in calb 1/2, but not another molecule of calb 2 (as in the calb 2 homodimer) since the second calb 2 does not bind C a 2 + . Apparently, salt affects the helicity of calb 1/2 greater than calb 2, since it binds two C a 2 + instead of one. This agrees with the fact that C a 2 + binding affinity has a strong electrostatic dependence. One may also speculate about point mutations of calbindin that have partly isolated the location of some important electrostatic interactions. A study that engineered all possible mutants of three surface charges E17, D19, and E26 to the corresponding neutral amides E17Q, D19N, and E26Q showed a reduction in -AGtotal of C a 2 + binding as an increasing number of surface charges are neutralized (Linse et al, 1988); however, the cooperativity increased. Linse et al., proposed that the neutralization of surface charges diminishes the electrostatic repulsion of negative charges, allowing a closer approach of the two C a 2 + binding sites and facilitating the transmission of conformational effects on C a 2 + binding to one of these sites. If we assume that salt neutralizes surfaces charges of calb 1 and calb 2, our results would appear to be consistent with this explanation. It is possible that cooperativity in calb 1/2 is maintained, or even increased, but the specific binding constants of the individual sites are drastically reduced to a much larger extent, and hence our result of 0 s ait - 0no salt = 9400 deg 48 cm 2/dmol for calb 1/2. On the other hand, Linse et al. (1991a) also observed that K + binds non-specifically to C a 2 + saturated calbindin D9k without changes in the protein conformation. This does not appear to be the case in our calb 1/2. If it were true, we would expect to observe the helicity of the calb 1/2 to be comparable to calb 2 under excess C a 2 + conditions in 0.1 M KC1, as was the case in the absence of salt. The conclusion of Linse et al. (1988), which could very well be true for calbindin D9k, does not appear to be applicable for our system. 4.8 Summary: C a 2 + Binding in 0.1 M K C 1 Our results in 0.1 M KC1 with calb 1, calb 2 and calb 1/2 are consistent with the following conclusions: 1) Calb 1 requires the presence of Ca 2 +»calb 2 in order to bind C a 2 + . Since calb 1 cannot bind C a 2 + on its own, the only salt effect was a decrease in helicity on apo-calb 1. 2) As was observed in the absence of salt, calb 2 binds C a 2 + independently of calb 1. The results in salt are still consistent with the proposed binding mechanism of calb 2 in the absence of salt, in which one equivalent of C a 2 + is sufficient to enable two calb 2 molecules to fold into moderate helicity. It is reasonable to speculate that salt has a dramatic effect on the binding of the first equivalent of C a 2 + (by screening important electrostatic interactions), but not on the C a 2 + affinity of the second calb 2 since it does not bind. Helicity in excess C a 2 + for calb 2 homodimer is decreased by A6 = 6200 deg cm2/dmol in 0.1 M KC1 compared to the absence of salt. 3) Calb 1/2 binds C a 2 + . The heterodimer forms, either in concert or stepwise, only in the presence of C a 2 + unlike the case in the absence of salt. Cooperativity and C a 2 + affinity are still present in 0.1 M KC1, but to a lesser degree. Salt affects the binding of C a 2 + by screening important electrostatic interactions in calb 2, which are then transmitted to calb 1. In excess 49 C a 2 + , an overall decrease in helicity of 9400 deg cm2/dmol is observed when compared with the absence of salt. 4) Our results are consistent with molecular dynamics and Monte Carlo studies (Kesvatera et al., 1994; Svensson et al., 1990) in which a dielectric equal to water is invoked for the interior of the protein. However, it is inconsistent with a mechanism in which non-specific K + binding on the surface of our fragments causes no significant change in conformation as proposed by Linse etal. (1988). The reaction equations for the binding mechanisms are summarized in Figure 17 below. Ca +2 calb 1 _ **" no C a + 2 binding Ca +2 calb 2 calb 2 Ca + 2»calb 2 Ca+ 2«(calb 2) 2 +2 i) Ca ii) calb 1 calb 2 calb l«(Ca + 2«calb 2) Ca +2 (Ca + 2»calb l)«(Ca + 2«calb 2) Figure 17. Reaction equations of C a 2 + binding of calb 1, calb 2 and calb 1/2 in 0.1 M KC1. Square brackets around the intermediate species indicate that it may be formed stepwise or in concert with C a 2 + binding. 50 5 Conclusion 5.1 Summary of Results We have synthesized and individually characterized the C a 2 + binding of the two EF-hands of calbindin Dcjk. Our results indicate that calb 1, which corresponds to the pseudo EF-hand (site I) of calbindin D9k, does not bind C a 2 + except in the presence of Ca 2+ bound calb 2, the archetypical EF-hand (site II). This effect is attributed to the positive cooperativity conferred by Ca 2 + , calb 2 to apo-calb 1. Calb 2 binds C a 2 + independent of calb 1. In the case of the calb 2, one equivalent of C a 2 + is sufficient to enable two calb 2 units to associate and fold into helical conformation. Salt has dramatic effects on the helicity and binding properties of calb 1 and calb 2. Helicity is reduced both in the presence and absence of C a 2 + for all three peptides. We attribute this to the screening of important electrostatic interactions that are required for C a 2 + to be bound. As expected, salt has a larger effect on calb 2 than calb 1, although the effect is transmitted by virtue of cooperativity as demonstrated by calb 1/2. 5.2 Future Studies The following are some suggestions for future studies of calb 1, calb 2 and calb 1/2: 1) Our current data provides indirect evidence for the mechanisms of C a 2 + binding in calb 2 and calb 1/2 (Figures 13, 14, and 16). Direct evidence should be determined to characterize the mechanism more fully. N M R could determine the nature of association of these two peptides as has been done by Kay et al. (1991) and Shaw et al. (1992b) for troponin C fragments. The stoichiometry would be known and a reliable C a 2 + binding constant could then be determined by C D and corroborated by N M R . Sedimentation equilibrium ultracentrifugation could also help determine the size of the associated species. One would expect that results will be consistent with the present data from AVand calorimetry. 51 2) Mutations could be done to affect the C a 2 + binding of the peptide fragments. Linse et al. (1994) engineered the mutant protein E60D and found that the protein has reduced C a 2 + affinity but enhanced cooperativity. A similar mutation could be done on calb 2 to determine whether this result is retained in the peptide fragment. Calb 1 could be mutated from the pseudo EF-hand to the archetypical EF-hand. This is likely to alter the coordination pattern for C a 2 + binding, as has been observed in the intact protein (Johansson et al., 1991). This would help address the issue of the role of the different sequence in the pseudo EF-hand as compared to the archetypical EF-hand. Mutations that alter the hydrophobic faces of calb 1 and calb 2 could determine their role in association, cooperativity and C a 2 + binding. 3) The linker between the two EF-hands of the intact protein could be incorporated into one or both fragments. The results of site I fragments, calb 1, CP-1, and F l , suggest an important contribution from the linker to C a 2 + binding and/or cooperativity. This may help address the role of the linker in cooperativity in the intact protein. 52 References Adler, A . J., Greenfield, N . J., & Fasman, G. D. (1973) Methods Enzymol. 27, 675-735. Ahlstrdm, P., Teleman, O., Kordel, J., Forsen, S., & Jdnsson, B . (1989) Biochemistry 28, 3205-3211. Akke, M . , Forsen, S., & Chazin, W. J. (1991) J. Mol. Biol. 220, 173-189. Akke, M . , Skelton, N . J., Kordel, J., Palmer, A . G. I l l , & Chazin, W. J. (1993) Biochemistry 32, 9832-9844. Brandts, J. F., & Kaplan, L . J. (1973) Biochemistry 12, 2011-2024. Brodin, P., Johansson, C , Forsen, S., Drakenberg, T., & Grundstrbm, T. (1990) J. Biol. Chem. 265, 11125-11130. Carlstrom, G., & W. J. Chazin (1993) J. Mol. Biol. 231, 415-430. Cheung, W. Y . , ed. (1980) Calcium and Cell Function. Academic Press: New York. Chiba, K. , Ohyashiki, T., & Mohri, T. (1983) J. Biochem. 93, 487-493. Christakos, S., Gabrielides, C , & Rhoten, W. B . (1989) Endocr. Rev. 10, 3-26. Dorrington, K . J., Kells, D. I. C , Hitchman, A . J. W., Harrison, J. E., & Hofmann, T. (1978) Can. J. Chem. 56, 492-499. Drakenberg, T., Hofmann, T. & Chazin, W. J. (1989) Biochemistry 28, 5946-5954. Finn, B . E. , Kordel, J., Thulin, E., Sellers, P., & Forsen, S. (1992) FEBS Lett. 298, 211-214. Foffani, M . T., Battistutta, R., Calderan, A . , Ruzza, P., Borin, G., & Peggion, E. , (1991) Biopolymers 31, 671-681. 53 Gillies, G. T., & Kupke, D. W. (1988) Rev. Sci. lustrum. 59, 307-313. Greenfield, N . , & Fasman, G. D. (1969) Biochemistry 8, 4108-4116. Herzberg, O., & James, M . N . G. (1985) Biochemistry 24, 5298-5302. Heizmann, C. W., ed. (1991) Novel Calcium-Binding Proteins: Fundamentals and Clinical Implications. Springer-Verlag: Berlin. Heizmann, C. W., & Hunziker, W. (1991) Trends Biochem. Sci. 16, 98-103 Hughson, F. M . , Barrick, D., & Baldwin, R. L . (1991) Biochemistry 30, 4113-4118. Johansson, C., Brodin, P., Grundstrom, T., Forsen, S., & Drakenberg, T. (1991) Eur. J. Biochem. 202, 1283-1290. Johansson, C., Ullner, M . , & Drakenberg, T. (1993) Biochemistry 32, 8429-8438. Kawasaki, H . , & Kretsinger, R. H . (1994) Calcium-Binding Proteins 1: EF-Hands. Academic Press: London. Kay, L . W., Forman-Kay, J. D., McCubbin, W. D. & Kay, C. M . (1991) Biochemistry 30, 4323-4333. Kesvatera, T., Jonsson, B. , Thulin, E., & Linse, S. (1994) Biochemistry 33, 14170-14176. Kordel, J., Forsen, S., & Chazin, W. J. (1989) Biochemistry 28, 7065-7074. Kordel, J., Forsen, S., Drakenberg, T., & Chazin, W. J. (1990) Biochemistry 29, 4400-4409. Kordel, J., Skelton, N . J., Akke, M . , & Chazin, W. J. (1993) J. Moi. Biol 231, 711-734. Kretsinger, R. H. , & Nockolds, C. E. (1973) / . Biol Chem. 248, 3313-3326. Kufty, R. M . , & Kretsinger, R. H . (1975) Science 187, 167-169. 54 Kupke, D. W., & Fox, J. W. (1989) Biochemistry 28, 4409-4415. Leavis, P. C , Rosenfeld, S. S., Gergely, J., Grabarek, Z., & Drabikowski, W. (1978) J. Biol. Chem. 253, 5452-5459. Levine, B . A. , & Williams, R. J. P. (1980) In Calcium and Cell Function (Cheung, W. Y . , ed.) p 32, Vol II. Academic Press: New York. Linse, S., Brodin, P., Drakenberg, T., Thulin, E., Sellers, P., Elmden, K. , Grundstrom, T., & Forsen, S. (1987) Biochemistry 26, 6723-6735. Linse, S., Brodin, P., Johansson, C , Thulin, E., Grundstrom, T., & Forsen, S. (1988) Nature 335, 651-652. Linse, S., Johansson, C , Brodin, P., Grundstrom, T., Drakenberg, T., & Forsen, S. (1991a) Biochemistry 30, 154-162. Linse, S., Helmersson, A. , & Forsen, S. (1991b) J. Biol. Chem. 266, 8050-5054. Linse, S., Thulin, E., & Sellers, P. (1993) Protein Sci. 2, 985-1000. Linse, S., Bylsma, N . R., Drakenberg, T., Sellers, P., Forsen, S., Thulin, E., Svensson, L . A. , Zajtzeva, I., Zajtsev, V . , & Marek, J. (1994) Biochemistry 33, 12478-12486. Linse, S., & Forsen, S. (1995) Adv. Second Messenger Phosphoprotein Res. 30, 89-152. Ngai, S-M., & Hodges, R. S., (1992) / . Biol. Chem. 267, 15715-15720. Nakayama, S., & Kretsinger, R. H. (1994) Annu. Rev. Biophys. Biomol. Struct. 23, 473-507. Newsholme, P., Angelos, K. L. , & Walsh, D. A . (1992) J. Biol. Chem. 267, 810-818. Osterhout, J. J., Handel, T., Na, G., Toumadje, A. , Long, R. C , Connolly, P. J., Hoch, J. C , Johnson, W. C , Live, D., & DeGrado, W. F. (1993) J. Am. Chem. Soc. 114, 331-337. 55 Permyakov, E. A. , Medvekin, V . N . , Mitin, Y . V. , & Kretsinger, R. H . (1991) Biochem. Biophys. Acta 1976, 67-70. Reid, R. (1987) Biochemistry 26, 6070-6073. Ringer, S. (1883a) J. Physiol. 4, 29-42. Ringer, S. (1883b) J. Physiol. 4, 222-225. Schachter, D. (1980) Fat Soluble Vitamins (Deluca, H . F., & Suttie, J. W., eds.). University of Wisconsin Press: Madison, WI. Scholtz, J. M . , Marqusee, S., Baldwin, R. L. , York, E. L . , Stewart, J. M . , Santoro, M . , & Bolen, D. W. (1991) Proc. Natl. Acad. Sci. USA 88, 2854-2858. Seamon, K. B . & Kretsinger, R. H . (1983) In Calcium in Biology (Spiro, T. G., ed.). John Wiley and Sons: New York. Shaw, G. S., Hodges, R. S., & Sykes, B . D. (1990) Science 249, 280-283. Shaw, G. S., Golden, L . F., Hodges, R. S., & Sykes, B . D (1991a) J. Am. Chem. Soc. 775,5557-5563. Shaw, G. S., Hodges, R. S., & Sykes, B. D., (1991b) Biochemistry 30, 8339-8347. Shaw, G. S., Findlay, W. A. , Semchuk, P. D., Hodges, R. S., & Sykes, B . D. (1992a) J. Am. Chem. Soc: 114, 6258-6259. Shaw, G. S., Hodges, R. S., & Sykes, B . D. (1992b) Biochemistry 31, 9572-9580. Shoemaker, K . R., Kim, P. S., York, E. J., Stewart, J. M . , & Baldwin, R. L . (1987) Nature 326, 563-567. Skelton, N . J., Kordel, J., Forsen, S., & Chazin, W. J. (1990a) / . Moi. Biol. 213, 593-598. 56 Skelton, N . J., Forsen, S., & Chazin, W. J. (1990b) Biochemistry 29, 5752-5761. Skelton, N . J., Kordel, J., Akke, M . , & Chazin, W. J. (1992a) J. Mol. Biol. 227, 1101-1117. Skelton, N . J., Akke, M . , Kordel, J., Thulin, E., Foren, S., & Chazin, W. J. (1992b) FEBS Lett. 303, 136-140. Svensson, B. , Jonsson, B. , & Woodward, C. E. (1990) Biophys. Chem. 38, 179-183. Svensson, B. , Jonsson, B. , Woodward, C. E., & Linse, S. (1991) Biochemistry 30, 5209-5217. Szebenyi, D. M . E., & Moffat, K. (1986) J. Biol. Chem. 261, 8761-8777. Tsuji, T., & Kaiser, E. T. (1991) Proteins 9, 12-22. Vogel, H . J., Drakenberg, T., Forsen., S., O'Neil , J. D. J., & Hofmann, T. (1985) Biochemistry 24, 3870-3876. Wada, A . (1976) Adv. Biophys. 9, 1-63. Wesolowski, T. A. , Boguta, G., & Bierzynski, A . (1990) Protein Eng. 4, 121-124. Williams, T. C , Corson, D. C , Oikawa, K. , McCubbin, W. D., Kay, C. M . , & Sykes, B . D. (1986) Biochemistry 25, 1835-1846. 57 Chapter 2: The a,cc-Corner Super-Secondary Structure Motif as Protein Folding Initiator The only solid piece of scientific truth about which I feel totally confident is that we are profoundly ignorant about nature. Lewis Thomas 58 1 Introduction1 1.1 The Protein Folding Problem How does a protein fold into its native structure? This question has been the subject of intense research since the pioneering days of denaturing proteins with acid and heat early this century. At that time, only various gross properties such as changes in viscosity and coagulation were observed. A better understanding of the denaturation process had to wait until the covalent structure of proteins was known. Sanger's work during the 1950's on the linear sequence of proteins layed the foundation for the discovery of more complex and higher order protein structures. Soon, secondary structures such as a-helices and (3-sheets were found (Pauling & Corey, 1951). This was followed quickly by the discovery of tertiary structures, that is, the arrangement of secondary structures in space. With this knowledge in hand, Christian Anfinsen (1973) studied the denaturation/renaturation process of a small protein ribonuclease. He showed that chemically denatured ribonuclease refolds to its native form after the denaturant was removed. The significance of this observation is that the information needed to specify the complex three-dimensional structure of ribonuclease is contained in its amino acid sequence alone. This was confirmed by Merrifield, who chemically synthesized the amino acid sequence of ribonuclease and found that it spontaneously folded into the active form (Gutte & Merrifield, 1971). Anfinsen's and Merrifield's work provided a well accepted principle in the field of protein folding: the amino acid sequence of a protein is the sole determinant of the native structure. A corollary of this principle is that the protein does not fold randomly, that is, it does not sample all the possible conformations before reaching the native structure (Levinthal, 1968). A simple calculation shows that the folding pathway cannot be random. A protein of 100 amino acids with two conformations available to each residue can have 1 0 3 0 total 1 Portions of this chapter have been previously published: Tsai, F. C. S., & Sherman, J. C. (1993) Biochem. Biophys. Res. Comm. 196 (1), 435-439. 59 conformations. If the protein required 10~12 s to sample each conformation, it would take 30 billion years, longer than the age of the universe, to reach its native form. Since proteins fold within seconds or minutes, it is clear that the folding pathways are specific and directed, and not random. The objective of researchers in the protein folding field is to determine the laws that govern how a protein folds and to predict the protein structure from its amino acid sequence. To reach this objective, two criteria are required (Fersht, 1993): 1) thermodynamic and structural characterization of the folded and unfolded protein, and any intermediates in the folding pathway, and 2) kinetic characterization of transition states and intermediates (see Figure 18). Folded State An ensemble of conformations Transition States with weak local structure and fluctuations. Figure 18. Criteria required for solving the protein folding process. A l l species in the pathway must be structurally and kinetically characterized. Reproduced from Fersht, 1993. Fulfilling these criteria would allow one to characterize, and therefore, determine the mechanism of the entire folding reaction. However, in spite of the apparently narrow criteria required, the protein folding problem is still inextricable and labyrinthine. Several models have been proposed in attempting to fulfill the criteria described above. Some theories overlap and have similarities with others, while some are completely exclusive. 60 It would be useful to be briefly acquainted with these models so that experimental work may be regarded contextually. The models have been conveniently categorized into kinetic and structural models (Kim & Baldwin, 1982, 1990). Kinetic models are aimed at determining the factors that control the rate of folding without being too specific about structures. Kinetic models include the following: 1) The biased random search model, where the number of possible conformations are drastically reduced by using only sterically possible conformations. Levitt, the main proponent, suggested that rapid folding may be achieved when the backbone conformation is sufficiently similar to that of the native structure (Levitt & Warshel, 1975; Levitt, 1976). At this point, the major free energy barrier (loss of entropy of the polypeptide chain) has been overcome. This result, combined with rapid non-specific collapse suggests that the biased random search may be a viable model in the early stages of folding. 2) The nucleation-growth model (Kim & Baldwin, 1982), where an initial fold guides subsequent steps, either sequentially and accumulating intermediates, or rapidly without intermediates. 3) The diffusion-collision model (Karplus & Weaver, 1976, 1994), where short segments of polypeptide chains fold independently into microdomains. These microdomains are unstable, but they diffuse, collide, coalesce and become stable. Karplus and Weaver have suggested that a series of these steps could lead to the native structure. 4) Sequential folding model (Cook et al, 1979; Kim & Baldwin, 1980), where folding occurs in a unique and definite sequence of steps, analogous to a metabolic pathway. Structural models are concerned with the actual structures of intermediates as well as their order in the folding pathway. Current structural models include: 1) The framework model (Ptitsyn, 1973, 1987), where hydrogen bonded secondary structures are formed early in the folding pathway and are then "locked" into their tertiary structure. The secondary, tertiary and quaternary structure of the protein folds in successive stages. 61 2) The modular assembly model (Kim & Baldwin, 1982), which proposes that a small protein folds by parts, and that subdomains could be important structures as folding intermediates. Each part of the protein can fold at different times, but complete folding of any one part occurs essentially at once. 3) The molten globule model (Kuwajima, 1989; Goldberg et al, 1990), which refers to a liquid-like and fluctuating ("molten") structure while still being compact and having a high content of secondary structure ("globule"). The molten globule is proposed to be the state of the protein just before folding into the native structure. The work done on these models still excludes the prodigious work done on prediction methods (Fasman, 1989), such as those based on hydrophobicity (Rose & Wolfenden, 1993; Ponnuswamy, 1993), the stability of proteins (Murphy & Freire, 1992), packing interactions (Richards & L im, 1994), neural network predictions (Benner & Gerloff, 1993), local independent nucleated units of structure (Srinivasan & Rose, 1995), and lattice statistical mechanics (Dill, 1985). Examples and detailed discussions are available for these models and the reader is referred to them for more comprehensive reviews. 1.2 The oc,a-Corner Motif as Protein Folding Initiator The various theories provide useful paradigms within which one can empirically examine the way a protein folds. However, the work described in this chapter by no means attempts to prove or disprove any one theory or another. It is an attempt to contribute to the empirical database from which one can improve and test theories in the future. We hope that our results will allow one to predict and alter folding in a variety of proteins. We have chosen to study the a,a-corner motif for this purpose. Formation of local secondary structure or hydrophobic clusters has been suggested as a plausible first step in protein folding (Wright et ah, 1988). The study of structured peptide 62 fragments as protein folding initiators is conducive for testing theories such as the framework or nucleation-growth model. The premise of these theories is that local secondary structures restrict conformational space and hence influence early folding events prior to the formation of super-secondary or tertiary structures. Recent observations of short helical peptides, reverse turns and hydrophobic clusters have supported this theory (Dyson & Wright, 1991). We will examine the a,a-corner motif as a folding initiator and a possible intermediate of the protein folding pathway. The oc,cc-corner was first identified by A. V . Efimov (1984a, b, 1994) as a common secondary structural motif in over 20 proteins. It consists of two a-helices linked by a connection (a short loop of two to eight amino acids) and packed at approximately right angles to one another. Although the left and right-handed mirror image forms are possible, only the right-handed (clockwise) chain course in the region of the connection, has been observed (see Figure 19). A left-handed a,a-corner would require that one of the helices turns by 180°, which Efimov shows to be sterically prohibited. The length of the connection between the two helices is restricted to a minimum of two amino acid residues due to the steric interaction of the amino acid residues of the two a-helices. Longer connections are known that consist of an extended P-strand, an a-helical fragment, a p-bend or a combination of such regions. Regardless of the length of the connection, Efimov has found that only the right-handed a,a-corner exists. The (|> and \|/ angles1 of the residues around the connection give the following sequence profile: OCR-OCR-CCL-P-P-OCR-OCR, where OCR, CCL, and P are residues with right-handed oc-helix, 1 Definition of $ and \|/ angles: \\f refers to rotations about the C a - C single bond; <}) refers to rotations about the C a - N single bond. R: H 63 a left-handed a-helix and |3-strand conformation, respectively. The sequence preferences are also generalized into hydrophobic and hydrophilic clusters along the a,a-corner motif. Efimov consistently found that the last hydrophobic residue of the first helix and the first hydrophobic residue of the second helix form a hydrophobic pair in positions 1 and 8 (see Figure 20). N-terminus N-terminus Left-handed a,oc-Corner Right-handed a,a-Corner Counter-clockwise Chain Course Clockwise Chain Course Figure 19. Schematic diagram of an a,a-corner. Only the right-handed cc,a-corner has been found in proteins. Cylinders represent a-helices. From Efimov, 1984a, b. 64 Figure 20. Schematic representation of a right-handed a,cc-corner with a short connection consisting of two peptide units. Hydrophobic side chains are designated by solid circles; hydrophilic side chains and Gly are shown by open circles or not shown at all; side chains which are partly immersed in the hydrophobic core are shown as semi-solid circles and such side chains can be either hydrophobic or hydrophilic; see also text. Reproduced from Efimov, 1984a. 65 In other words, two hydrophobic residues that are eight residues apart in the sequence form what Efimov calls the "necessary cluster" of hydrophobic residues. There is also a necessary cluster between the last hydrophobic residue of the first helix and the first residue of the p conformation connection, ie. three residues apart. Overall, Efimov characterizes the a,oc-corner with the following features: 1) hydrophobic residues in a 1-3-8 sequence to form necessary clusters between the first and second helices; 2) residues Gly, Arg or Lys in the last position of the first helix and 3) hydrophilic or small residues (Gly, Ala , Pro) in the first position of the second helix. The following amino acid preferences were also found. The last residue of the first helix avoids steric hinderances with the connection by being in the ctL conformation. As such, the most common amino acid for this position is Gly or Ala, or a residue with a long and flexible side chain such as Lys or Arg. The (XL residue cannot be Val , He, Leu, Phe, Tyr, Trp or Thr. Efimov's survey of 16 proteins shows 11 Gly, 3 Lys, one Arg, one His and no large hydrophobic residues at this position. The first residue in the second helix was found to be invariably in the P conformation with its side chain directed towards the second helix axis. Under these conditions, large hydrophobic residues are prohibited in this position. In 33 sequences, Efimov found no Val, He, Leu, Phe nor Trp in the first position of the second helix; Tyr occurs twice. More often, Ser, Thr, Asn or Asp residues occupy this position, which also permits hydrogen bonding from the side chain to the free N H groups of the backbone. This feature was found to be in every a-helix with a p conformation in the first position. Others have already shown the preference for hydrophilic residues in the first position of oc-helices for the same reasons (Finkelstein, 1976). As a protein folding initiator, Efimov analyzed the packing of other secondary structures around the a,cc-corner motif. He found that successive addition of cc-helices around the a,a-corner motif afforded known super-secondary structures of non-homologous proteins. 66 He was able to formulate a "structural tree" of proteins containing oc,a-corners, while the root from which all structures emerge is the oc,a-corner itself. Efimov suggested that the a,a-corner is the nucleus in the process of the folding of a,a-corner-containing proteins. 1.3 The a,a-Corner from the a-Chain of Horse Methemoglobin If the a,oc-corner motif is a protein folding initiator, a synthetic a,a-corner peptide would exhibit helicity independent of the remainder of the protein. We chose the a,a-corner of the a-chain of horse methemoglobin (Figures 21 and 22) to test this hypothesis. The sequence comprises the F and G helices (residues 80 to 108) of methemoglobin (Efimov, 1984a, b), which is structurally a tetrameric analogue of myoglobin. The folding of myoglobin has been described in the framework model, in which individual a-helices form independently of each other but are stabilized by helix pairing reactions, so that a stable framework of helical structure is formed which acts as a hydrophobic container for the heme group (Hughson & Baldwin, 1989). The discovery of a protein folding initiator in hemoglobin would help affirm the current folding pathway proposed for myoglobin. The sequence of our peptide, called aac-heme2 , is as follows: 80 108 Ac-L-S-D-L-S-N-L-H-A-H-K -L -R # -V*-D*-P-V-N -F -K-L-L-S-H -C -L-L-S-T-NH 2 1 3 8 The residues in bold (L91, V93, and F98) are the hydrophobic residues of Efimov's necessary cluster. The residue marked # is in the aL conformation in the native protein and residues 2 The "aac" denotes a,a-corner and "heme" denotes that it is the sequence of the a,a-corner from hemoglobin. 67 marked * are in the (3 conformation. The N-terminus was acetylated and the C-terminus amidated to avoid the introduction of charges that are not present in the native protein. We obtained CD spectra of aac-heme at various concentrations in aqueous buffer and various amounts of trifluoroethanol (TFE) to test the potential of this motif as a folding initiator. W37/R.O 042/12.04 Figure 21. Amino acid sequence of the oc-chain of horse methemoglobin (Reproduced from Ladner et al, 1977). 68 Figure 22. Schematic structure of hemoglobin indicating the location of the F and G helices (Reproduced from Imai, 1982). 2 Experimental 69 A l l reagents were used without further purification. Peptide aac-heme was synthesized by solid phase methodology using Fmoc chemistry on an A B I 431A automated peptide synthesizer using FastMoc 3 chemistry protocal. Activation was accomplished with 2-( lH-benzotriazol-l-yl)-l, 1, 3, 3-tetramethyluronium hexafluorophosphate (HBTU) and 1-hydroxybenzotriazole (HOBt) in the presence of diisopropylethylamine (DIPEA). Fmoc protected amino acid derivatives, solvents and other reagents were purchased from Advanced ChemTech (Louisville, Kentucky). The peptide was purified by reverse phase H P L C on a Phenomenex Cjg column (25 x 1 cm or 25 x 2.25 cm) using a water/acetonitrile gradient with 0.1% trifluoroacetic acid. The peptide was determined to be greater than 98% pure by analytical reverse phase HPLC and characterized by liquid secondary ion mass spectrometry: m/z 3312 (M+H)+ calcd 3312.90. Circular dichroism spectra were acquired on a Jasco J-710 spectropolarimeter equipped with Neslab RTE-210 circulating water bath. A l l stock solutions, buffers and C D samples were filtered through a 0.45 urn filter prior to data acquisition. A l l spectra were acquired at 4° C after a 10 minute equilibration. A l l spectra were acquired as an average of three scans using a cuvette with a 1 mm path length. Concentrations were determined from amino acid analysis using norleucine as an internal reference. Other conditions are specified in the individual figures. Error bars represent one standard deviation from the average of three determinations. 3 User Bulletin No. 33, November 1990, Applied Biosystems, Foster City, C A . 70 3 Results A C D spectrum of aac-heme in aqueous solution is shown in Figure 23. We determined the helicity of aac-heme by analysis of its CD spectra. Helicity is often measured using equation 1: (Oobs " 6o)/(9max ~0o) 1 where 90bs is the observed ellipticity at 222 nm, 6 0 is the ellipticity at 222 nm of the fully denatured peptide and 9max is the ellipticity at 222 nm for the peptide when it is fully helical (Lyu et al., 1991a). We determined 0 O (0 + 500 deg , cm 2 »dmol- 1 ) by measuring 6222 in the presence of 6 M guanidine hydrochloride. We determined ©max by measuring 6222 in the presence of 60% trifluoroethanol (TFE). A CD spectrum of aac-heme in 60% TFE is shown in Figure 24. The maximum at 195 nm and minima at 208 and 222 nm are characteristic of a-helices. The observed ratio of these ellipticities is typical of a fully helical peptide (Lyu et al., 1991b; Manning, 1989). Furthermore, a plot of 6222 versus TFE concentration (Figure 25) shows that ellipticity levels off at about 50% TFE. Thus, we take 0 m a x to be -36,000 ± 3000 deg ,cm 2»dmol" 1 . This is a typical value for ©222 for a fully helical peptide of length 29 residues in TFE (Lyu et al., 1991b). Figure 26 shows a plot of 0222 versus concentration of aac-heme in aqueous solution. From the data in Figure 26, equation 1 shows that the helicity of aac-heme in aqueous solution increases from 19% to 37% over the concentration range of 13-93 u M . Figure 23. CD spectrum of aac-heme (93 mM) in 50 mM sodium borate, 0.1 M NaCl, pH 7,4° C. 8 is in deg cm2/dmol. 72 Figure 24. Circular dichroism spectrum of aac-heme (13 |iM) in 60% TFE, 4°C. 6 is in deg cm2/dmol. 73 40000 35000 30000 25000 —\ o E CM E o D) CD 20000 CM CM CM CO 4-* CD 15000 10000 5000 Figure 25. Helicity of aac-heme as a function of TFE percentage. Error bars represent one standard deviation from the average of four determinations. (13 [iM aac-heme, 4°C) 74 15000 o E T3 CJ E o D) CO T3, CM CM CM CO CO 10000 5000 H 20 40 60 80 Concentration of aac-heme, uM 100 Figure 26. Helicity of aac-heme as a function of concentration. Error bars represent one standard deviation from the average of two determinations. (50 m M sodium borate, pH 7,0.1 M NaCl, 4°C) 75 4 Discussion The a-chain of horse methemoglobin is tetrameric. Myoglobin, which is the monomeric analogue of hemoglobin, is much more widely studied in the field of protein folding. Therefore, the discussion will occasionally refer to the much more available work done on myoglobin. The topology of a hemoglobin subunit and myoglobin is similar enough to accept this is a reasonable approach (Fermi & Perutz, 1981; Fermi et al, 1984). 4.1 Concentration Studies Our synthetic peptide, aac-heme, represents a protein fragment that corresponds to the a,a-corner of the a-chain of horse methemoglobin. This peptide exhibits moderate helicity in aqueous solution and this helicity is concentration dependent. Concentration dependence of helicity is often an indication of aggregation. This is particularly common in amphiphilic peptides (Osterhout et ah, 1993) or in protein fragments (Goodman & Kim, 1989), which are devoid of the tertiary packing interactions that they encounter in the intact protein. The peptide aggregates to compensate for the absence of tertiary interactions. Thus, the aggregation of aac-heme in aqueous solution is not surprising since aac-heme represents a portion of hemoglobin that normally holds many tertiary packing interactions with the remainder of the protein. In myoglobin, helices F and G comprise three of the six principal contacts between two helices, and constitute about half of the hydrophobic contribution to the free energy of association (Richmonds & Richards, 1978). Other myoglobin helices, which have been proposed to be important intermediates in the folding reaction, have also exhibited concentration dependent helicity. N M R and C D experiments indicate that peptides corresponding to the G and H helices of sperm whale myoglobin aggregate in solution above 0.5 to 1 m M concentration (Hughson et al, 1991; Waltho et al, 1993; Shin et al, 1993b). 76 Tertiary interactions have been implicated in the stability of these helices during the folding reaction of myoglobin (Waltho et al., 1993; Shin et al, 1993b). Likewise, the concentration dependent C D spectra of aac-heme indicate that tertiary interactions are important for its stability. The significance for folding initiation will be discussed below. 4.2 TFE Studies In the hydrophobic TFE solutions (Storrs et al., 1992; Nelson & Kallenbach, 1986), aac-heme is fully helical. The moderate helicity of this a,oc-corner peptide in water alone and the high helicity in the hydrophobic TFE solutions suggest that this oc,a-corner sequence is capable of helical stability independent of the remainder of the protein and that the helicity is enhanced by a hydrophobic environment. Recent investigations have indicated that TFE induces helical conformation in a peptide sequence only in regions for which there is a significant preference for helical conformations; in the absence of such a preference, no evidence for secondary structure formation, helix or otherwise, has been found. Sonnichsen et al. (1992) found that a 28 residue peptide of actin is highly helical except in the region of a P-turn. Shin et al. (1993a) found that a P-turn in the peptide corresponding to the G-H turn of myoglobin remains intact in high TFE aqueous solutions while the sequence flanking the turn is in helical conformation. The investigators suggest that these non-helical regions, even in TFE, act as a helix stop signal. Thus, it is possible in aac-heme that the P conformations in the connection are present even under high TFE concentrations. N M R experiments would be amenable for determining whether this is the case. At present, the results indicate that the sequence has an inherent propensity for helicity, independent of the remainder of the protein. 77 4.3 Protein Folding Initiation Proteins fold rapidly from the denatured state, usually within seconds. Exceptions occur when folding involves proline cis-trans isomerism (Brandts et al, 1975; Schmid, 1992) or rearrangements of disulfide bonds (Creighton, 1988). N M R experiments indicate that secondary structures stabilized by tertiary interactions are formed within 10 ms as shown by N M R pulse labeling experiments (Udgaonkar & Baldwin, 1988, 1990). The formation of secondary structures occurs on the submicrosecond timescale. Experiments on weakly folded protein segments also suggest that a hydrophobic collapse that withdraws many nonpolar side chains from the solvent precedes secondary structure formation (Dill, 1985; Matthews, 1993). This has been suggested to provide hydrophobic docking locations for amphiphilic helices or stabilizing P-sheet hydrogen bonds in a hydrophobic core (Matthews, 1993). The results with aac-heme are consistent with a folding mechanism for the native hemoglobin whereby a hydrophobic collapse occurs first, followed by the formation of secondary structure. Our results suggest that the a,a-corner could well initiate the secondary structural formation in hemoglobin and that this initiation would be even more pronounced if it was preceded by a hydrophobic collapse of the nonpolar side chains in hemoglobin. The propensity of aac-heme to be moderately helical in aqueous solution and highly helical in TFE is consistent with this mechanism. The concentration dependent helicity indicates the importance of tertiary interactions that are required once the helices are formed. Although aac-heme is moderately helical in aqueous solution, we have no evidence yet that the peptide adopts an a,oc-corner motif. It is possible that the peptide adopts a helical conformation along its entire length. This scenario suggests that an isolated oc,a-corner is unstable without the interactions that are present in the intact protein. A metastable structure does not necessarily preclude its role in folding initiation since it could help prevent the locking in of incorrectly folded regions at the early stages of folding (Wright et al., 1988). Local structures that initiate folding may be retained and further stabilized or rearranged into diffferent 78 structures as the protein folds into its tertiary conformation; that is, non-native structures may be significantly populated along the folding reaction (for examples, see Chaffotte et al., 1992; Thomas et al., 1990). Thus, initiation structures direct the pathway(s) of folding by restricting conformational space and these structures may not persist in the final folded state (Wright et al, 1988). If aac-heme is helical along its entire length, it can still be a protein initiator but the a,a-corner motif is unstable in the absence of other interactions. The concentration dependence of helicity in aac-heme would suggest tertiary interactions as a likely source for further stabilization. Since Sonnichsen et al. (1992) and Shin et al. (1993a) observed breaks in the helical conformation in regions of a peptide in TFE that are not helical in the intact protein, it is possible that we may observe the oc,a-corner under similar conditions. This scenario suggests that the isolated cc,a-corner is stable and a super-secondary structure could initiate and direct the pathway(s) of hemoglobin folding as Efimov predicted. 4.4 Protein Folding Models A l l protein folding models have theorized on how the folding reaction might begin. We have shown that aac-heme may initiate the folding of intact hemoglobin by having observed its moderate helicity in aqueous solution. How does aac-heme apply to various models of the protein folding reaction? Let's consider a few that have been applied to globin proteins. The folding of myoglobin has been described using the framework model (Baldwin, 1991). The framework model (Wright et al., 1988) proposes the formation of local secondary structure or hydrophobic clusters in several regions of the polypeptide chain to be the initial step in the protein folding reaction. Several short peptides of natural sequences have been observed to adopt secondary structure that could advance this theory (Goodman & Kim, 1989; Oas & Kim, 1988; Dyson et al, 1988a, b; Shoemaker et al, 1987a, b). In the intact protein, 79 these structures are only marginally stable, transient structures in rapid dynamic equilibrium with the fully unfolded states and are presumed to be very similar to the folded structures observed in short peptide fragments. They are stabilized by short range and medium range (three to four residues) interactions determined by the local amino acid sequence. Our observations with aac-heme are compatible with this model. The moderate helicity of aac-heme in aqueous solution indicates that it is moderately stable and may resemble the native structure. Helicity in TFE increases, indicating that aac-heme is capable of helical stability independent of the remainder of the protein. The diffusion-collision model has also been used and predicted the formation of helix F and G (ie. aac-heme) to be part of one of the first two important clusters of secondary structures formed along the folding reaction (Bashford et al., 1988). Other theories, which consider the packing interactions of a-helices, have also been proposed (Cohen et al., 1979; Richmond & Richards, 1978). Ptitsyn & Rashin (1975) showed that helices F, G and H were also found to be favourable as possible "crystallization centres" of folding. Clearly, aac-heme has been associated with important aspects of protein folding models. 80 5 Conclusion 5.1 Summary of Results We have synthesized and characterized by C D a peptide, called aac-heme, corresponding to the a,a-corner of the a-chain of horse methemoglobin. Aac-heme is moderately helical in aqueous solution and helicity is concentration dependent; it is highly helical in TFE solutions. Our results indicate that aac-heme may be a protein folding initiator in the folding reaction of methemoglobin, as Efimov predicted. This initiation could be enhanced by a hydrophobic environment. At present, model studies and experimental data appear to implicate aac-heme as an important element in the folding of globin proteins. More experimental data is necessary to ascertain the relevance of a,oc-corners as protein folding initiators, in globin proteins or other proteins. 5.2 Future Studies The following are some suggestions for future work on aac-heme or similar systems: 1) Since the helicity of aac-heme has been verified, determination of the exact conformation of each residue would be an important step in characterizing the a,oc-corner sequence. N M R experiments would be most amenable to this end. Two results are likely: the sequence is entirely helical, or there is a break where the sequence is in (3-conformation in the native protein. The implications for these possibilities have been discussed. 2) Mutations of the sequence would affect the stability of aac-heme. This may lead to conclusions about how each residue functions to provide stability to the secondary or super-secondary structure. One could also test for Efimov's proposal of a "necessary cluster" of hydrophobic residues in the sequence by replacing them with hydrophilic residues. 81 3) Computer modeling of the sequence would provide a theoretical framework with which to compare empirical data. Since prediction programs are widely available, one may be able to identify other a,a-corner motifs in proteins, or design one de novo. 4) Efimov provided a survey of proteins containing oc,a-corners. We have only tested one of these. Synthesizing and characterizing other sequences of a,a-corners could determine if this motif has common or disimilar properties, ie. helicity, stability, and conformation. 82 References Anfinsen, C . B . (1973) Science 181, 223-230. . Baldwin, R. L . (1991) Ciba Found. Symp. 161, 190-205. Bashford, D., Cohen, F. E., Karplus, M . , Kuntz, I. D., & Weaver, D. L . (1988) Proteins 4, 211-227. Benner, S. A. , & Gerloff, D. L . (1993) FEBS Lett. 325, 29-33. Brandts, J. F., Halvorson, H . R., & Brennan, M . (1975) Biochemistry 14, 4953-4963. Chaffotte, A . F., Cadieux, C , Guillou, Y. , & Goldberg, M . E. (1992) Biochemistry 31, 4303-4308. Cohen, F. E. , Richmond, T. J., & Richards, F. M . (1979) J. Moi. Biol. 132, 275-288. Cook, K. H. , Schmid., F. X . , & Baldwin, R. L . (1979) Proc. Natl. Acad. Sci. USA 76, 6157-6161. Creighton, T. E. (1988) Proc. Natl. Acad. Sci. USA 85, 5082-5086. Di l l , K, A . (1985) Biochemistry 24, 1501-1509. Dyson, H . J., Ranee, M . , Houghton, R. A. , Lerner, R. A. , & Wright, P. E. (1988a) J. Moi. Biol. 201, 161-200. Dyson, H . J., Ranee, M . , Houghton, R. A. , Wright, P. E., & Lerner, R. A . (1988b) / . Moi. Biol. 201, 201-217. Dyson, H . J., & Wright, P. E. (1991) Annu. Rev. Biophys. Biophys. Chem. 20, 519-538. Efimov, A . V . (1984a) FEBS Lett. 166, 33-38. Efimov, A . V . (1984b) Moi. Biol. 18, 1239:1251. 83 Efimov, A . V . (1994) FEBS Lett. 355, 213-219. Fasman, G. D., ed. (1989) Prediction of Protein Structure and the Principles of Protein Conformation. New York: Plenum. Fermi, G., & Perutz, M . F. (1981) Atlas of Molecular Structures in Biology. 2. Haemoglobin and Myoglobin. Oxford: Claredon Press. Fermi, G., Perutz, M . F., Shaanan, B. , & Fourme, R. (1984) J. Mol. Biol. 175, 159-174. Fersht., A . R. (1993) FEBS Lett. 325, 5-16. Finkelstein, A . V . (1976) Mol. Biol. 507, 879-886. Goldberg., M . E., Semisotnov, G. V. , Friguet, B. , Kuwajima, K. , Ptitsyn, O. B . & Sugai, S. (1990) FEBS Lett. 263, 51-56. Goodman, E. M . , & Kim, P. S. (1989) Biochemistry 28, 4343-4347. Gutte, B. , & Merrifield, R. B. (1971) J. Biol. Chem. 246, 1922. Hughson, F. M . , & Baldwin, R. L . (1989) Biochemistry 28, 4415-4422. Hughson, F. M . , Barrick, D., & Baldwin, R. L . (1991) Biochemistry 30, 4113-4118. Imai, K. (1982) Allosteric effects in haemoglobin. Cambridge, U K : Cambridge University Press. Karplus, M . , & Weaver, D. L . (1979) Biopolymers 18, 1421-1437. Karplus, M . , & Weaver, D. L . (1994) Protein Sci. 3, 650-668. Kim., P. S., & Baldwin, R. L . (1980) Biochemistry 19, 6124-6129. Kim. , P. S., & Baldwin, R. L . (1982) Ann. Rev. Biochem. 51, 459-489. 84 Kim. , P. S., & Baldwin, R. L . (1990) Ann. Rev. Biochem. 59, 631-660. Kuwajima, K. (1989) Proteins 6, 87-103. Ladner, R. C , Heidner, E. J., & Perutz, M . F. (1977) / . Mol. Biol. 114, 385-414. Levinthal, C. (1968) J. Chim. Phys. 65, 44-45. Levitt, M . (1976) J. Mol. Biol. 104, 59-107. Levitt, M . , & Warshel, A . (1975) Nature 253, 694-698. Lyu, P. C , Sherman, J. C , Chen, A. , & Kallenbach, N . R. (1991a) Proc. Natl. Acad. Sci. 88, 5317-5320. Lyu, P. C , Wang, P. C , Liff, M . I., & Kallenbach, N . R. (1991b) J. Am. Chem. Soc. 113, 3568-3572. Manning, M . C. (1989) J. Pharm. Biomed. Anal. 7, 1103-1119. Matthews, C. R. (1993) Annu. Rev. Biochem. 62, 653-683. Murphy, K. P., & Freire, E. (1992) Adv. Prot. Chem. 43, 313-361. Nelson, J. W., & Kallenbach, N . R. (1986) Proteins 1, 211-217. Oas, T. G., & Kim, P. S. (1988) Nature 336, 43-48. Osterhout, J. J., Handel, T., Na, G., Toumadje, A. , Long, R. C , Connolly, P. J., Hoch, J. C , Johnson, W. C , Live, D., & DeGrado, W. F. (1993) / . Am. Chem. Soc. 114, 331-337. Pauling, L . & Corey, R. B. (1951) Proc. Natl. Acad. Sci. USA 37, 729-740. Ptitsyn, O. B. , & Rashin, A . A . (1975) Biophys. Chem. 3, 1-20. 85 PonnusWamy, P. K . (1993) Prog. Biophys. Molec. Biol. 59, 57-103 Ptitsyn, O. B . (1973) Doklady Akad. Nauk. SSSR 210, 1213-1215. Ptitsyn, O. B . (1987) J. Protein Chem. 6, 273-293. Richards, F. M . , & Lim, W. A . (1994) Quart. Rev. Biophys. 26, 423-498. Richmond, T. J., & Richards, F. M . (1978) J. Moi. Biol. 119, 537-555. Rose, G. D., & Wolfenden, R. (1993) Annu. Rev. Biophys. Biomol. Struct. 22, 381-415. Schmid, F. X . (1992) In Protein Folding (Creighton, T. E. ed.) pp 197-241. New York: W. H. Freeman and Company. Shin, H - C , Merutka, G., Waltho, J. P., Wright, P. E., & Dyson, H . J. (1993a) Biochemistry 32, 6348-6355. Shin, H - C , Merutka, G., Waltho, J. P., Tennant, L . L . , Dyson, H . J., & Wright, P. E. (1993b) Biochemistry 32, 6356-6364. Shoemaker, K. R., Kim, P. S., York, E. J., Stewart, J. M . , & Baldwin, R. L . (1987a) Nature 326,562-561. Shoemaker, K . R., Fairman, R., Kim, P. S., York, E. J., Stewart, J. M . , & Baldwin, R. L . (1987b) Cold Spring Harbor Symp. Quant. Biol. Ul, 391-398. Sonnichsen, F. D., Van Eyk, J. E., Hodges, R. S., & Sykes, B . D. (1992) Biochemistry 31, 8790-8798. Srinivasan, R., & Rose, G. D. (1995) Proteins 22, 81-99. Storrs, R. W., Truckses, D., & Wemmer, D. E. (1992) Biopolymers 32, 1695-1702. Thomas, G. J. Jr., Becka, R., Sargent, D., Yu , M . - H . , & King, J. (1990) Biochemistry 29, 4181-4187. 86 Udgaonkar, J. B. , & Baldwin, R. L . (1988) Nature 335, 694-699. Udgaonkar, J. B . , & Baldwin, R. L . (1990) Proc. Natl. Acad. Sci. USA 87, 8197-8201. Waltho, J. P., Feher, V . A. , Merutka, G., Dyson, H . J., & Wright, P. E. (1993) Biochemistry 32, 6337-6347. Wright, P. E., Dyson, H . J., & Lerner, R. A . (1988) Biochemistry 27, 7167-7175. 

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:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0061741/manifest

Comment

Related Items