UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Expression and characterization of two recombinant mammalian metalloproteins : |b bovine microsomal… Funk, Walter David 1990

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

Item Metadata


831-UBC_1991_A1 F86.pdf [ 11.07MB ]
JSON: 831-1.0100483.json
JSON-LD: 831-1.0100483-ld.json
RDF/XML (Pretty): 831-1.0100483-rdf.xml
RDF/JSON: 831-1.0100483-rdf.json
Turtle: 831-1.0100483-turtle.txt
N-Triples: 831-1.0100483-rdf-ntriples.txt
Original Record: 831-1.0100483-source.json
Full Text

Full Text

EXPRESSION AND CHARACTERIZATION OF TWO RECOMBINANT MAMMALIAN METALLOPROTEINS: BOVINE MICROSOMAL CYTOCHROME b5 AND HUMAN SERUM TRANSFERRIN (N LOBE) By Walter David Funk B.Sc. (Hon) 1984 University of British Columbia A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF Ph.D. in THE FACULTY OF MEDICINE DEPARTMENT OF BIOCHEMISTRY We accept this thesis as conforming to the required standard The University of British Columbia November, 1990 (c) W A L T E R D. F U N K , 1990 In presenting this thesis in partial fulfillment 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 of Biochemistry The University of British Columbia 2075 Westbrook Place Vancouver, Canada V6T 1W5 August 1, 1990 Abstract Two separate systems were developed for the expression of recombinant metalloproteins. A synthetic gene encoding the lipase-solubilized form of bovine liver microsomal cytochrome 65 was designed and assembled for expression in E. coli. Analysis of the initial recombinant cytochrome revealed differences in several physical characteristics of the molecule compared to the authentic bovine liver species, including a reduction potential that was lower by 17 mV. Further studies showed the primary sequence of the initial recombinant differed from the authentic protein in the amidation status of three residues which, when corrected yielded a recombinant protein identical in behaviour to the authentic protein. The participation of Ser64 in the stabilization of the oxidized form of cytochrome 65 was investigated using site-directed mutagenesis to alter this residue to Ala, which was predicted to ablate a hydrogen bond formed between the protein and heme-propionate 7. Spectroelectrochemical analysis of this variant showed that the reduction potential had been shifted downwards by 7 mV, in contrast to predictions from a structural model describing the red/ox behaviour of cytochrome £5 (Argos and Mathews, 1975). The role of heme carboxylates in determining the reduction potential was confirmed for both the wild-type and Ala64 variants by heme replacement studies using the esterified derivative of protoporphyrin IX, suggesting that the presence of free carboxylates contributes to the stabilization of the oxidized species. In addition, constructions for the expression of the trypsin-solubilized form of bovine liver microsomal cytochrome 65 and the erythrocytic form of human cytochrome 65 are described. A tissue culture cell system was developed for the expression of the N-terminal half molecule of human serum transferrin. The recombinant molecule (hTF/2N) was secreted at high levels from selected eukaryotic cells, and displayed high identity with i i i the proteolytically-derived molecule from authentic human serum transferrin as judged by sequence analysis, electrophoretic mobility and iron binding capacity. A construction for the expression of the C-terminal half molecule was assembled but failed to express recombinant protein when introduced into tissue culture cells. The production of these two heterologous expression systems allows for high-level recovery of recombinant protein and provides a convenient approach to structure-function studies employing site-directed mutagenesis techniques. i v T A B L E OF C O N T E N T S Abstract ii Table of Contents iv List of Tables viii List of Figures ix List of Abbreviations xi Acknowledgements xiv I. Introduction 1 A . Cytochrome b$ 3 1. Historical 4 2. Biosynthesis and Physiological Role 6 3. Structural Considerations 11 i) Isolation of Soluble and Membrane-Binding Forms 11 ii) Amino Acid Sequence Analysis 13 iii) Structural Analysis 16 4. Nucleotide Sequence Analysis 25 B. The Transferrins 27 1. Historical 28 2. Physiological Role and Expression 31 3. Fundamental Properties 34 i) Physiochemical and Conformational Characteristics 34 ii) Carbohydrate Composition 35 iii) Amino Acid Sequence Analysis 36 iv) Metal Binding Characteristics 37 v) The Two Lobes of Transferrin 41 4. Nucleotide Sequence Analysis 43 V i) Chicken Ovotransferrin 43 ii) Human Serum Transferrin and Lactoferrin 5. Molecular Structure of Transferrin and Lactoferrin i) Human Lactoferrin ii) Rabbit Serum Transferrin 6. Iron Binding Site and Structural Considerations II. Experimental Procedures and Materials A General Materials and Supplies B. Cytochrome 65 Studies 1. Gene Construction 2. Host Selection 3. Induction Studies 4. Protein Purification 5. Dimethylesterheme IX Substitution 6. Tryptic Peptide Mapping 7. Amino Acid Sequence Determination 8. Limited Tryptic Hydrolysis 9. Site-directed Mutagenesis 10. Spectroelectrochemistry C. Human Serum Transferrin Studies 1. Isolation of Human Serum Transferrin cDNA 2. Expression Vector Assembly 3. D N A Transfection and Eukaryotic Cell Culture 4. Immunoprecipitation and Western Blotting 5. Protein Purification 46 48 48 50 53 57 57 57 57 58 59 59 60 61 62 62 62 63 63 63 64 65 66 66 v i 6. Amino-terminal Sequence Analysis 67 7. h T F / 2 c - p N U T Assembly 68 III. Results 69 A . Cytochrome £5 Studies 69 1. Gene Design and Synthesis 69 2. Expression Studies 73 3. Amino-terminal Sequence Analysis 77 4. F P L C Analysis of Authentic and Recombinant Cytochromes 65 77 5. Tryptic Peptide Mapping 79 6. Primary Sequence Correction 83 7. Constructions for the Expression of Trypsin-solubilized and Human Erythrocytic Cytochrome 65 85 8. Spectroelectrochemical Studies 89 B. Transferrin Studies 93 1. c D N A Isolation and Prokaryotic Expression Constructions 93 2. Eukaryotic Expression Construction 98 3. Isolation and Characterization of Recombinant h T F / 2 N 100 4. h T F / 2 C Studies 105 IV. Discussion 109 A . Prokaryotic Expression Studies 109 B. Analysis of the Q11,E13,D57 Triple Mutant of Cytochrome 65 113 C. S64A Substitution and D M E Replacement Studies 115 D. Expression of hTF/2N in Eukaryotic Cells 118 E . h T F / 2 C Construction 120 F. Additional Projects and Work in Progress 121 V . References 123 v i i VI. Appendix I: Measurement of the Reduction Potential of Biological Molecules v i i i TABLES Table I: Amino-terminal Sequence Analysis of Recombinant Lpb$ Table II: H P L C Retention Times of Tryptic Peptides from Recombinant Bovine Microsomal Lipase-solubilized b$ (Q11,E13,D57) and Authentic Bovine Microsomal Trypsin-solubilized Cytochrome 65 Table III: Midpoint Reduction Potentials of Recombinant Cytochrome 65 Species Table IV: Amino-terminal Sequence of Human Transferrin and of the Recombinant hTF/2N Molecule i x FIGURES Figure 1: Biochemical Pathways That Involve Cytochrome 65 as a Mediator of Electron Transfer 9 Figure 2: Isolated Forms of Bovine Cytochrome 65 12 Figure 3: Primary Sequence Comparison of Selected Hepatic and Erythrocytic Cytochrome 65 Species 15 Figure 4: Ribbon-representation of the Crystal Structure of ' Recombinant Lipase-solubilized Cytochrome £5 17 Figure 5: Ser64 Hydrogen-bonding Network 20 Figure 6: Alternate Heme Orientations in Cytochrome 65 21 Figure 7: A Model for the Reduction/Oxidation Equilibrium of Cytochrome 65 23 Figure 8: Structure of Ferriprotoporphyrin IX 24 Figure 9: Structure of the Glycan Chains of Human and Rabbit Serum Transferrin 36 Figure 10: Amino Acid Sequence Identity Between Transferrin Family Members 38 Figure 11: Possible Scheme for the Evolution of the Present Human Serum Transferrin Gene 45 Figure 12: Sequence Comparison of the Four Domains of Ovotransferrin 47 Figure 13: Metal Ligands of Transferrin 50 Figure 14: Overall Molecular Organization of Rabbit Serum Transferrin 51 Figure 15: Structural Organization of the N-terminal Lobe of Rabbit Serum Transferrin 52 Figure 16: Region of the N lobe Encoded by Exon 2 54 Figure 17: Assembly Strategy for the Cytochrome 65 Gene Cassette 70 Figure 18: Second Strand Fill-in Reaction of B1-B4 Oligonucleotides 71 Figure 19: Nucleotide Sequence of the Lipase-solubilized, Bovine Microsomal Cytochrome 65 Gene Cassette 73 X Figure 20: Transcriptional Fusion Strategy for the Expression df Recombinant Cytochrome 65 74 Figure 21: E. coli Host Selection for Cytochrome 65 Expression 75 Figure 22: IPTG Induction Time Course of Lpb$ Expression 76 Figure 23: F P L C Elution Profiles of Recombinant and Authentic Microsomal Cytochrome 65 80 Figure 24: F P L C Elution Profiles of the Limited Digestion Products of Recombinant Lpb$ Species (Q11,E13 and D57) 85 Figure 25: PCR Mutagenesis Strategy 86 Figure 26: Thin Layer Spectra of Recombinant Wild-type and S64A Forms of Lpb5 90 Figure 27: In vitro Mutagenesis for the Assembly of hTF/2N Coding Sequences 94 Figure 28: Vectors for the Expression of Recombinant hTF/2N in bacteria 97 Figure 29: p N U T - h T F / 2 N Expression Vector Assembly 99 Figure 30: Western Blot Analysis of Immunoprecipitates from Transfected Eukaryotic Cell Cultures 101 Figure 31: Isolation and Analysis of Recombinant hTF/2N 102 Figure 32: Visible Absorption of Recombinant hTF/2N and Human Serum Transferrin 104 Figure 33: Iron Titration of the Apo Form of Recombinant h T F / 2 N 106 Figure 34: Assembly Strategy for Recombinant hTF/2C Coding Sequence 107 Figure 35: Comparison of the Upstream Sequences from Various Expression Vector Constructions 111 Figure 36: Relative Positioning of Asp57 in the Crystal Structure of LpZ?5 114 Figure 37: Comparison of the Crystal Structures of Recombinant and Authentic Cytochrome 65 at Positions 11 and 13 116 x i Abbreviations Used A Absorbance A Angstrom A L A Aminolevulinic acid A T P Adenosine triphosphate B H K Baby Hamster Kidney b.p. before present bp base pair C D . Circular dichroism c D N A complementary deoxyribonucleic acid CoA Coenzyme A Da Dalton D H F R Dihydrofolate reductase D M E Protoporphyrin IX dimethylester D M E M Dulbecco's Modified Essential Medium D N A Deoxyribonucleic acid E. coli Escherichia coli E D T A Ethylenediamine tetraacetic acid EPR Electron paramagnetic resonance Er&5 Erythrocytic form of cytochrome 65 ExoIII Exonuclease III F P L C Fast Protein Liquid Chromatography g Universal gravitational constant h G H Human growth hormone H P L C High Pressure Liquid Chromatography IPTG isopropylthiogalactoside IRE Iron responsive element x i i Association constant kbp kilobase pair kDa kiloDalton Klenow large fragment of E. coli D N A Polymerase I L F Lactoferrin Lpb5 Lipase-solubilized form of hepatic microsomal cytochrome 65 m R N A messenger ribonucleic acid mV millivolt melanoTF melanotransferrin (p97) M T metallothionein M T X methotrexate NadChol sodium deoxycholate NaDodS04 sodium dodecylsulphate N A D H nicotinamide adenine dinucleotide (reduced) N A D P H nicotinamide adenine dinucleotide phosphate (reduced) N M R Nuclear magnetic resonance N T A Nitrilotriacetic acid O.D. Optical density OMb5 Mitochondrial Outer Membrane Cytochrome 65 O T F Ovotransferrin O T T L E Optically Transparent Thin Layer Electrode P A G E Polyacrylamide gel electrophoresis PBS Phosphate buffered saline PCR Polymerase chain reaction PMSF Phenylmethylsulfonyl fluoride R N A Ribonucleic acid r R N A ribosomal Ribonucleic acid x i i i SCE Standard Calomel Electrode SHE Standard Hydrogen Electrode SSC Standard Saline Citrate SV40 Simian Virus 40 T F Transferrin T F A Trifluoroacetic acid T F / 2 C Carboxy-terminal half molecule of transferrin T F / 2 N Amino-terminal half molecule of transferrin Tp Trypsin T P C K L-l-tosylamide-2-phenyl-ethylchloromethyl ketone Tris Tris(hydroxymethyl)aminomethane xiv A C K N O W L E D G E M E N T S I'd like to thank all the little people I mean the really large and important people with whom I have had the pleasure of working with over the past five years. This project involved literally dozens of people and I deeply appreciate the attention and patience awarded to me by co-workers. In particular, huge thanks to Bob Woodworth and Anne Mason for their achievements on the transferrin project and Marcia Mauk and Paul Barker for their guidance through the realms of protein chemistry. The spectacular graphics in this dissertation were produced by the lightning keystrokes of the Brayer lab members; thanks very much to Terrance Lo, Michael Murphy and Albert Berghius for great screenplay. As a longstanding member of the MacGillivray lab, I've seen many come and go (and some come back again). Very special thanks to Marlys Koschinsky for sharing the excitement and comradery of science and to Bruce Ritchie and Dave Banfield for much needed laughs. I'd like to wish each and every lab member (past, present and future) a cheery sink full. Having two research supervisors has been the best of two worlds. Thank you Grant for sharing with me the beauty of precision. The lesson may not yet be completed, but will never be forgotten. Ross; you have given me opportunities when I really needed them and have shown me not only great science, but even greater friendship. Thanks forever. Pamela; your patience, encouragement and understanding mean everything. Thank you for letting me dream. 1 I. INTRODUCTION The past decade has seen the development of powerful methods for the expression of recombinant proteins in heterologous systems. Typically, these systems rely on the introduction and overexpression of cloned D N A molecules in either bacterial or tissue culture hosts and often allow the recovery of large amounts of protein that may normally occur with extremely low natural abundance. Combining these techniques with the ability to manipulate D N A molecules precisely through the process of in vitro mutagenesis (Zoller and Smith, 1983) has revolutionized the manner in which the structural and functional properties of proteins are studied. Non-genetic approaches have been limited not only by the natural abundance and availability of a given protein but also by the accuracy and specificity of chemical modification techniques that have been commonly used to probe protein structures. Genetic manipulations in concert with the capacity to synthesize extended-length oligonucleotides chemically now allows for the assembly of virtually any imaginable coding sequence, such that the term "synthetic biology" has now been coined to describe these new approaches. This thesis reports the development and initial characterization of systems for the expression of recombinant forms of two mammalian iron-containing proteins. Cytochrome 65 is a small (16 kDa) heme protein found in the microsomal fraction and on the inner mitochondrial membrane of liver, where it acts as an essential participant in a wide variety of electron transfer reactions. In both hepatic forms, this cytochrome exists as a soluble heme-binding core attached by a short linker sequence to a membrane-anchoring domain. Serum transferrin (78 kDa) is the primary carrier of iron in plasma and as such is central to maintaining metal homeostasis. This protein is 2 composed of two highly similar lobes, each containing a single iron-binding center. Our initial attempts have been directed toward expression of these two proteins in forms that are amenable to purification and characterization. To this end, a bacterial expression system for the production of the soluble domain of bovine cytochrome 65 and a tissue culture expression system for production of the N-terminal lobe of human serum transferrin have been developed. A review of previous literature concerning both of these proteins is presented below to illustrate our interest in the investigation of both proteins through molecular genetic techniques. 3 A. Cytochrome D5 Cytochromes 65 are acidic cytochromes that possess a non-covalently-bound protoheme IX prosthetic group with histidyl residues providing the proximal and distal ligands to the heme iron. Cytochromes 65 are found in most mammalian tissues in the microsomal and mitochondrial fractions. The hepatic form was discovered first and has been studied most thoroughly. This protein plays a central role in reduction/oxidation (red/ox) processes at the endoplasmic reticulum (Oshino,1982) and can be described in terms of three structural domains. The water soluble heme-binding domain consists of approximately 88 amino acid residues and contains an abundance of acidic residues that are concentrated near the edge of the heme-binding cleft. A short linker peptide of approximately 10 residues joins the soluble domain to a membrane-anchoring C-terminal domain of approximately 40 predominantly hydrophobic residues. The existence of other forms of cytochrome 65 in mammalian cells has been conclusively demonstrated only in the past decade. Erythrocytic cytochrome 65 was described as a soluble form of the hepatic molecule which was originally thought to be derived by proteolytic cleavage of the linker peptide during maturation of the red blood cell (Passon et al.,\912). Recent analysis of the amino acid (Kimura,1984) and m R N A sequence (Steggles et al., 1989) of this form shows it to differ by one amino acid residue from the soluble domain of its hepatic counterpart leading to speculation of the existence of at least separate exons, if not separate genes, for the two forms. Nisimoto (1977) and Ito (1980) purified a third member of the cytochrome 65 family from the outer membrane of mitochondria, OM65. Amino acid sequence analysis of OM65 (Lederer et al., 1983) has shown it to share approximately 58% sequence identity with microsomal 4 cytochrome 65. No reports concerning the molecular genetics of this protein have appeared. In the following review, the term "cytochrome 65", or its abbreviation "65", will apply to the microsomal cytochrome unless otherwise indicated. 1. Historical The initial observation of non-hemoglobin pigmented substances in tissues was made by MacMunn in the late 19th century. With little support from the scientific community of the time, it remained for Keilin in 1925 to "re-discover" these coloured proteins, which he subsequently termed cytochromes, signifying coloured pigments (Keilin, 1925). Keilin described four primary absorption bands for this pigment, the A band (615nm-593nm), B band (567.5nm-561nm), C band (554.4nm-546 nm) and D band (532nm-51 lnm) and went on to demonstrate that the cytochrome absorption pattern was due to the presence of three distinct components, which he termed a',b' and c'. Considerable variability was observed in the B band of cytochromes from different organisms and tissues, leading to the identification and definition of cytochrome b (Keilin,1925), cytochrome b\ (Keilin, 1925), cytochrome bi (Bach et al., 1946), cytochrome 63 (Martin and Morton, 1955), and cytochrome 64 (Egami et a/.,1953). Each member of the 6-type family was characterized by activity towards red-ox agents and spectroscopic behavior, with each member containing a protoheme or related non-formyl heme non-covalently bound by the protein (Wainio,1982). The first description of what has since been commonly termed cytochrome 65 was made by Keilin and Hartree (1940) who described a 6-type cytochrome in pig kidney, which they called b\. Later, Yoshikawa (1951) reported a similar protein in 5 liver, calling it cytochrome b\ while Strittmatter and Ball (1954) referred to a similar enzyme from rat liver microsomes as cytochrome m. Cytochrome b$ was initially used as a descriptor for a protein from Ceropia silkworm larvae (Pappenheimer and Williams, 1954), though it now appears likely that the silkworm cytochrome was actually cytochrome c\ of the respiratory chain (Wainio,1982). Chance and Williams (1954) recognized the spectroscopic similarity between cytochrome m and the silkworm protein and renamed the microsomal protein cytochrome 65. Microsomal b$ was subsequently shown to be present in the liver of guinea pig, rabbit and calf and could be released from microsomal preparations with a pancreatic lipase preparation (Strittmatter and Velick,1956). The same study showed £5 to be distinct from mitochondrial b\ and identified the prosthetic group subsequently as iron protoporphyrin IX. Detergent solubilization of microsomal membranes allowed identification of the hydrophobic C-terminal membrane-binding domain (Ito and Sato, 1968). Passon and co-workers (1972) identified a soluble form of cytochrome b$ in mammalian erythrocytes. The purified cytochrome was shown to participate in the methemoglobin reductase system and was proposed to arise from hydrolytic cleavage of a membrane-bound precursor during the maturation of the erythrocyte. However, recent evidence from nucleotide sequence analysis of erythrocytic 65 m R N A has shown conclusively that this protein is distinct from the hepatic species and is translated from a different m R N A (see IIB4, Nucleotide Sequence Analysis). The existence of a mitochondrial species of 65 was first proposed by Raw and Mahler (1959), who identified an NADH-cytochrome c reductase activity in the outer membrane. However several years passed before this outer membrane protein, OM65, 6 was purified and shown to be unequivocally distinct from the microsomal species (Nisimoto et al, 1977; Ito,1980). 2. Biosynthesis and Physiological Role Descriptions of b-type cytochrome in the microsomal fraction of hepatocytes were reported as early as 1954 (Chance and Williams). Cytochrome b$ has been detected immunologically on both the rough and smooth endoplasmic reticulum (Remacle et al.,\916), and a NADH-reduceable cytochrome was also reported in nuclear and Golgi membrane fractions (Ichikawa and Yamano, 1970; Fleischer et al., 1971). These findings suggested that cytochrome b$ may be distributed on most cellular membranes, with the notable exception of the plasma membrane (Fleischer et al., 1971). With the exception of heart and skeletal muscle, almost all tissues yield detectable amounts of cytochrome 65 in their microsomal fractions, though relative levels of the protein vary from 0.03 nmol/mg protein in the rat testis to 1.0 nmol/mg protein in the liver (see Oshino,1982). Mitochondrial 65 has been localized to the outer face of the outer membrane (Fukushima and Sato,1973) while erythrocytic £5 is strictly a cytosolic protein (Passon et a/.,1972). Cytochrome 65 is synthesized on free ribosomes (Rachubinski et al., 1980; Okada et al., 1982) after which the protein relocates to membrane surfaces where binding occurs. The ability of cytochrome £5 to insert spontaneously into many artificial and natural membrane systems (Rogers and Strittmatter,1975) suggests that the observed concentration of 65 on microsomal membranes in vivo may be dependent on transport or exchange processes as not yet described. Little is known concerning the mechanism of heme insertion, though the ability of apo-65 to assimilate heme spontaneously suggests that a specific heme "insertase" is not required. Apo-65 is easily hydrolyzed by 7 proteases, and the protein undergoes an increase in structural order upon binding heme (Huntley and Strittmatter, 1972). The half-life of hepatic 65 in vivo has been estimated to be approximately 100 hr. (Omura et al.,\961) while the heme moiety has a half life of just 55 hr. (Druyan et al., 1968), indicating that the heme group may undergo exchange reactions with other proteins. Hepatic cytochrome b$ plays a central role in the electron transfer systems of the microsomal membrane. In general, cytochrome 65 mediates one electron transfer reactions between two flavoproteins, NADH-cytochrome 65 reductase and N A D P H -cytochrome c reductase that participate in a variety of metabolic reactions as shown in Figure 1. Both reductase molecules share a similar organization with 65 in that they consist of a soluble catalytic domain which can be released from membranes by proteolysis (Spatz and Strittmatter, 1973). Both the hydrolytic and native forms of cytochrome 65 reductase are capable of reducing the hydrolytic and native forms of 65, and reactions between the native forms of the two proteins are accelerated under conditions which favour protein-protein interaction (Rogers and Strittmatter, 1973). The catalytic cycle of this enzyme involves the transfer of two electrons from the reductase to two molecules of 65 in sequential reaction steps (see Mathews and Czerwinski, 1976 for review). Reduction of 65 by NADPH-cytochrome 65 reductase requires the reductase to be in its native membrane-bound state (Bilimoria and Kamin,1973). However, the cytochrome P450 family appears to be the principal electron acceptor of cytochrome c reductase (Jansson and Schenkman,1977), thus the physiological significance of cytochrome c reductase in i n -dependent reactions remains unclear. 8 A ) NADH Fp N-Hydroxyamine reductase Fatty Acid Elongation enzyme 7 5 zx-Sterol .^ -desaturase Acyl CoA / \ -desaturase Acyl CoA -desaturase N^N. Acyl CoA -desaturase N< Phospholipid desaturase Alkyl acyl GPE desaturase NADPH CytP450 B) metMb reductase mb. CmetMb Mb Q metHb reductase erb metHb Hb 9 Figure 1: Metabolic Pathways That Involve Cytochrome D5 as a Mediator of Electron Transfer (modified from Oshino, 1982). Panel A: Liver microsomal cytochrome £5 (m^) mediates single electron transfers between membrane-bound NADH-cytochrome 65 reductase (Fppj) and several fatty acid desaturase systems, fatty acid elongation enzyme N-hydroxyamine reductase and the Cytochrome P450 family of enzymes (see Oshino, 1982). In vitro, cytochrome 65 may also shuttle electrons from NADPH-cytochrome c reductase (Fpj) to the same cytochrome P450 enzymes though the physiological significance of this observation is weak due to the high relative activity of Fpr_> and the abundance of N A D H in intact liver. Panel B: A n NADH-dependent reductase activity has been identified in the muscle tissues of several organisms (see Livingston et al., 1985 for example). This reductase activity is required to maintain myoglobin (Mb) in a reduced (ferrous) state in order to bind oxygen reversibly, and in vitro, this reaction is increased 100-fold by the presence of cytochrome 65. Panel C: Methemoglobin (MetHb) reductase is the term commonly used to describe the soluble form of NADH-cytochrome 65 reductase found in the erythrocyte cytoplasm. This reductase system maintains hemoglobin in the ferrous state to ensure cooperative oxygen binding and utilizes the soluble erythrocytic form of cytochrome £5 (erb$). The principal cytochrome 65 oxidase activities can be classified either as desaturases or as members of the cytochrome P450 family. Acyl CoA desaturase activities catalyze the A 9 - , A6- and A 5 - desaturation of long-chain fatty acids at the expense of molecular oxygen and NAD(P)H or ascorbate (Stoffel,1961; Marsh and James, 1962). The involvement of cytochrome 65 in these reactions has been inferred because £5 is the only microsomal protein that can be reduced by all three reductants (Holloway and Katz,1972) and, desaturation is specifically inhibited by the presence of anti-^5 antibodies (Jansson and Schenkman,1977). Other desaturase processes that probably require the participation of cytochrome b$ include alkylacylglycerophosphoryl-ethanolamine desaturase (Paltauf and Holasek,1973), phospholipid desaturase (Pugh and Kutes,1977) and A7-sterol A-5 desaturase (Reddy et al.,\911). 10 The cytochrome P450 family catalyzes the hydroxylation of a wide spectrum of drugs and xenobiotics and as such plays a central role in detoxification. Cytochrome 65 is required for the maximal activity of many P450 reactions, including the oxidation of chlorobenzene (Lu et a/.,1974) and 3,4-benzpyrene (West et al.,1974) and the demethylation of benzphetamine (Imai and Sato, 1977) and /?-nitroanisole (Sugiyama et al., 1979). In addition, hepatic cytochrome 65 has been implied as an intermediary in many other microsomal reactions that use NAD(P)H as a source of reducing equivalents. These processes include fatty acid elongation (Keyes et al.,\919) and N-hydroxyamine reduction (Kadlubar and Ziegler,1974). The soluble erythrocytic form of cytochrome 65 functions as a shuttle in the methemoglobin reductase system (Passon et al.,\912), transferring electrons from methemoglobin reductase to methemoglobin (see Figure 1). Hemoglobin must be maintained in the reduced state to bind oxygen and thus erythrocytic 65 is essential for normal respiration. Both the erythrocytic £5 reductase and erythrocytic £5 are soluble molecules, and initial studies suggested these molecules are produced from the hydrolysis of membrane-bound precursor forms during the maturation of erythrocytes (Slaughter and Hultquist,1979). Analysis of cDNA sequences for the erythrocytic forms of cytochrome 65 show that this molecule is translated from a distinct erythrocytic m R N A , with a predicted sequence slightly different from that of hepatic 65 (A. Steggles, personal communication). The existence of separate genes for the hepatic and erythrocytic forms of 65 may be pertinent to the condition of methemoglobinemia (Hegesh et al., 1986), in which low levels of erythrocytic 65 activity accompany high levels of methemoglobin. These patients may maintain normal hepatic 65 expression and function, while a genetic lesion specifically affects the expression of erythrocytic 65. 11 The existence of a cytochrome 65-like molecule in the outer membrane of the mitochondrion was first described by Raw and Mahler (1959) as a component of a rotenone-insensitive cytochrome c reductase system. The term outer membrane, or OM65 was used to describe this hemoprotein (Fukushima and Sato, 1972) and sequence analysis confirmed that OMA5 is closely related to the hepatic and erythrocytic 65 species (Lederer et al., 1983). The rat microsomal and mitochondrial proteins exhibit a 58% sequence identity. The basis for the localization of OM&5 in the mitochondrial membrane remains enigmatic because small differences in the sequences of membrane-binding domains are not known to provide membrane binding specificity. The presence of membrane-specific protein-protein interactions may be a factor in compartmentalizing membrane proteins (Okada et al, 1982). 3. Structural Considerations i) Isolation of Soluble and Membrane-Binding Forms The release of a soluble heme-binding domain of microsomal cytochrome 65 can be achieved with a variety of proteolytic activities. Pioneering work by Strittmatter and Velick (1956) used a pancreatic lipase preparation for this purpose. The active proteolytic agent in this preparation may be cathepsin-type proteases released during the isolation of microsomes. Kajihara and Hagihara (1968) used trypsin and subtilisin in their preparations. These proteins produced 65 derivatives with different N - and C-termini as a result of differences in protease substrate specificity (Figure 2). By comparison, erythrocytic forms of cytochrome b$ maintain their N-terminal residues and end at position 97. 12 Heme-Binding Domain Residues 1-88 Linker Sequence Membrane-Binding Domain residues 108-133 133 97 97 Membrane-Binding Holoprotein (Hepatic Microsomes) Detergent-Solubilized Form Erythrocytic Form Lipase-Solubilized Form (Hepatic Microsomes) 3D-structure determined 88 Trypsin-Solubilized Form (Hepatic Microsomes) Figure 2: Isolated Forms of Bovine Cytochrome D5 (modified from Abe et al.,1985) The hepatic form of cytochrome £5 is a microsomal protein consisting of a cytoplasmic heme-binding domain, joined to a membrane anchoring domain by a short linker sequence. The erythrocytic form maintains an intact heme-binding domain along with a portion of the linker peptide. Isolation of the soluble domain by using pancreatic lipase preparations results in the removal of four residues from the N-terminus while trypsin solubilization hydrolyzes the heme-binding domain at both the N - and C-termini. 13 Detergent solubilization of microsomal preparations with Triton X-100 releases the intact cytochrome (Spatz and Strittmatter, 1971). The functional properties of the catalytic domain are not affected by the presence of the non-polar domain, though aggregation of the intact protein into octamers occurs in the absence of detergent (Oshino, 1978; Waino,1978). The insertion of 65 into lipid bilayers has been shown to occur as a "loose" complex in which the protein can readily exchange between lipid vesicles. Membrane bound 65 is susceptible to C-terminal cleavage by carboxypeptidase Y (Enoch and Strittmatter, 1979; Leto et at., 1980; Takagaki et al., 1983). Reconstitution into liposomes in the presence of detergent produces a complex in which the C-terminus is protected from carboxypeptidase Y activity, and transfer between vesicles is not observed. This observation suggests formation of a "tight" complex in which the C-terminus is now found in the lumen of the vesicle under these conditions. Studies on the topology of 65 in physiological membranes indicates that the C-terminal hexapeptide can be released from microsomal preparations by the action of trypsin (Ozols,1989), in agreement with membrane-binding models in which the hydrophobic domain loops into the bilayer and re-emerges on the same cytosolic face. Recent studies in which the hydrophobic domain sequence of cytochrome 65 was fused to the C-terminus of /J-galactosidase and subsequently expressed in E. coli have demonstrated that this domain directs attached proteins to membrane surfaces (George et al., 1989), and therefore, probably maintains a conformation which is independent of the soluble domain of cytochrome 65. 14 1 10 20 30 * * * * Human(l)a A E Q S D E A V K Y Y T L E E I Q K H N H S K S T U L I L H Human(e)k A E ° S D K A V K Y Y T L E E I Q K H N H S K S T U L I L H Bovine(l) c A E E S S K A V K Y Y T L E E I Q K H N N S K s T W L I L H BovineCe)^ A E E S S K A V K Y Y T L E E I Q K H N N S K s T W L I L H Rat(l) e A E Q S D K E V K Y Y T L E E I E K H K D S D s T W V I L H Chicken(l) f G S S E A G G E A w R G R Y Y R L E E V Q K H N N D s T U 1 I y H 40 50 60 * # * * Human(I) H K V Y D L T K F L E E H P G G E E V L R E Q A G G D A T E N F E D Human(e) H K V Y D L T K F L E E H P G G E E V L R E Q A G G D A T E N F E D Bovine(I) Y K V Y D L T K f L E E H P G G E E V L R E Q A G G D A T E N F E D Bovine(e) Y K V Y D L T K F L E E H P G G E E V L R E Q A G G D A T E N F E D Rat(l) H K V Y D L T K F L E E H P G G E E V L R E Q A G G D A T E N F E D Chicken(l) H R I Y D i T K F L D E H P G G E E V L R E Q A G G D A T E N F E D 70 80 90 # * * * Human(I) V G H S T D A R E M S < T F I I G E L H P D D R P K L N K P P E T L Human(e) V G H S T D A R E M S K T F I I G E L H P D D T P K L N K P P E P Bovine(l) V G H S T D A R E L S K T F I I G E L H P D D R S K i T K P S E S 1 Bovine(e) V G H S T D A R E L S K T F I I G E L H P D D R S K i T K P s E S Rat(l) V G H S T E A R E L S K T Y I I G E L H P D D R S K I S K P s E T L Chicken(l) V G H S T D A R A L S E R F 1 I G E L H P D D R P K L Q K P A E T L 100 110 120 130 * * * * HumanCI) I T T i D S S S S W w T N U V I P A I S A y A V A L M Y R L Y M A E Bovine(l) I T T I D S N P S U u T N W L I P A I S A L F V A L I Y H L Y I S E Rat(l) I T T V E S N S S w w T N u V I P A I S A L V V A L M Y R L Y H A E Chicken(l) I T T V Q S N s S s u S N u V I P A I A A 1 i V A L H Y R S Y H S E apredicted from cDNA sequence; Yoo and Steggles,1988. ^protein sequence; Abe et al.,1985. cpredicted from cDNA sequence; Cristiano and Steggles,1989. •^protein sequence; Douglas and Hultquist,1978. eprotein sequence; Ozols and Heinemann,1982. fpredicted from cDNA sequence; Zhang and Somervilie,1988. *axial histidyl residues 15 Figure 3: Primary Sequence Comparison of Selected Hepatic and Erythrocytic Cytochrome D5 Species. Published primary sequences were compared and residues at variance with the majority of mammalian residues are underlined. ii) Amino Acid Sequence Analysis The amino acid sequences of both the hepatic and erythrocytic forms of cytochrome 65 have been determined for several species, using both chemical and c D N A sequencing techniques. The results of sequence analysis of several species presented in Figure 3 clearly show the extraordinary sequence conservation seen for this protein. Sequence identity between avian and mammalian species is approximately 75%, indicating that the protein must be under considerable evolutionary constraint. The divergence of these two lineages has been estimated to have occurred 200 million years ago (Doolittle and Feng, 1987). A total of 54 positions are invariant in all known microsomal species. Significant differences are seen primarily at the extreme N - and C-termini and in the hinge region connecting the hydrophilic and hydrophobic domains of the protein. Sequence analysis has established that the N-termini of most species are acetylated. Only the porcine and chicken have free amino termini (Ozols,1989). As the first 15 residues are identical for porcine, horse and rat cytochromes, the lack of acetylation of the porcine molecule is surprising. The estimated molecular weight of intact bovine liver microsomal 65 determined from its amino acid sequence is 15,233 daltons (including heme), while the lipase and trypsin solubilized forms of the same species have molecular weights of 11,250 (residues 5-97) and 10,082 daltons (residues 7-88) respectively. 16 The initial report of the sequence of the erythrocytic and hepatic forms of bovine 65 concluded the sequence to be identical over residues 1-97 (Slaughter et al., 1982). The authors suggested that the erythrocytic form is generated from a full-length precursor that is identical to the hepatic form. Lysosomal proteases released during the maturation of red blood cells were postulated as being responsible for this conversion. This model has since been modified by the sequence determination of the C-termini of erythrocytic 65 from human, pig (Kimura et al., 1984) and rabbit (Schafer and Hultquist,1983). Residue 97 was identified as Pro and Ser for the human and porcine erythrocytic forms respectively, while the corresponding liver forms of both species had Thr at this position. This led to speculation that the two forms were encoded by distinct transcripts, either as the result of alternate splicing of exons from a single gene, or from the presence of two separate genes. Recent evidence provided by Southern analysis of chicken genomic D N A favours the latter of these two models (see IA4, Nucleotide Sequence Analysis). The complete amino acid sequence of the heme-binding domain of rat liver outer mitochondrial membrane cytochrome 65 has also been determined (Lederer et al., 1983). This sequence differs from that of the microsomal protein at 38 of 91 residues. When the non-conservative changes are mapped to the crystal structure of the calf microsomal 65 (Mathews et al.,\912), most replacements are found at the surface of the molecule. iii) Structural Analysis The crystal structure of the lipase-solubilized bovine liver microsomal 65 was first solved by Mathews et al. (1971a) to a resolution of 2.8A and later refined to 2.0A by the same workers (1971b) (see Fig 4). The protein has an approximately cylindrical Calf Liver Cytochrome b5 S64 Figure 4: Ribbon-representation of the Crystal Structure of Recombinant Lipase-solubilized Cytochrome bg The crystal structure of a triple mutant form of recombinant bovine cytochrome £5 (Ql 1,E13,D57) is represented in ribbon fashion (Funk et al., 1990b). The three mutant side chains and Ser64 are shown as ball and stick projections. The soluble domain consists of two smaller subdomains; the heme-binding domain (residues 21-78) is formed by four adjacent a helices and a /?-pleated sheet, while a smaller hydrophobic domain (upper in this representation) is comprised of residues 1-20 and 79-88. shape, with a 35A length and 30A diameter. Two helical subdomains that are separated by a ^-pleated sheet can be identified. The larger subdomain (residues 21-78) holds the heme moiety which sits in a distinctive crevice formed by four helices that run in anti-parallel directions. The floor of the binding pocket is formed by the ,9-pleated sheet 18 which separates the two domains. The heme group is bound non-covalently to the protein, with the heme iron coordinated to histidine residues 39 and 63. Several non-covalent interactions with 11 hydrophobic amino acid residues stabilize the binding orientation of the heme group. Compared to other heme proteins such as hemoglobin, the structural rigidity of the tightly packed heme crevice accounts for the low reactivity of 65 towards binding of oxygen and carbon monoxide (Mathews and Czerwinski,1976). The heme-binding domain is highly acidic with acidic groups outnumbering basic groups by 20 to 9 (Mathews, 1985). The negatively charged residues are concentrated at the heme edge, where the estimated net charge is -10 (including the propionate groups). Complex formation between cytochrome 65 and several reductases and oxidases is strongly dependent upon electrostatic interactions between the negatively charged face of b5 and corresponding positively charged surfaces on the other component. Based on the manipulation of high-resolution crystal structures, models of the complexes of b$ with cytochrome c (Salemme,1976), methemoglobin (Poulos and Mauk,1983 ), myoglobin (Livingston et al., 1985), and cytochrome P 4 5 0 c a m (Stayton et al., 1989) have been proposed which feature charge complementarity at the protein interface between acidic residues of 65 and lysyl residues of the binding partner. In agreement with such modelling, formation of the cytochrome ^cytochrome £5 complex is dependent on pH and ionic strength (Mauk et al., 1982). Reduction of the negative charge on the binding face of £5 by protein engineering (Rodgers et al., 1988) or neutralization of the lysyl residues of cytochrome c by chemical modification (Stoneheurner et al., 1979) greatly reduces the stability of the complex. Although hemoglobin and cytochrome 65 share little sequence identity, the overall structural organization of their heme binding regions are quite similar (Rossman 19 and Argos,1975). Both molecules bind heme in a non-covalent manner, with the propionate groups extending from the opening of the heme crevice. In particular, the four a-helices that frame the heme moiety of 65 can be accurately superimposed onto similar helices in both the A and B chains of hemoglobin. Other proteins that share structural homology with the heme-binding domain of 65 include yeast flavocytochrome £>2 (Jacq and Lederer,1974) and mammalian sulfite oxidase (Cohen and Fridovich,1971). Upon proteolytic digestion, both proteins release a "core" fragment of approximately 11 kDa which binds heme non-covalently and shares similar spectroscopic properties with 65. Guiard and Lederer (1979) have compared the amino acid sequences of the heme-binding cores of the three proteins and have defined a common "£5 fold" domain which has been subsequently identified in a fourth protein, nitrate reductase from Neurospora crassa (Le and Lederer, 1983). Each fold contains two domains that are separated by a /3-pleated sheet. A total of 13 invariant residues have been identified, including two axial histidines to which the heme is coordinated. A plausible scheme for the evolution of these proteins has been proposed in which coding sequence for a primordial heme binding fragment has been recruited for fusion with other enzymatic coding sequences. The heme group is almost entirely buried within cytochrome 65 and only the propionate groups are significantly exposed to solvent (Stellwagen,1978). One propionate extends fully into the solvent while the other is only partially exposed, folding back into the protein to form hydrogen bonds with the hydroxyl group of Ser 64 and a main-chain amide (see Figure 5). Two conformations are available for the heme moiety, related by a rotation of 180° about the a-^ meso axis (Figure 6). Analysis of the crystal structure of oxidized 65 indicated that the two heme orientations occur with a ratio of 4:1 20 Figure 5: Ser64 Hydrogen-bonding Network The hydrogen-bonding pattern at residue 64 of the recombinant cytochrome 65 is virtually identical to that reported by Mathews et al. (1971b) for the authentic bovine liver molecule. Heme propionate 7 forms two hydrogen bonds with Ser64; one carboxylate oxygen bonds with the side chain hydroxyl while the other bonds to the main chain amide of Ser64. The bond lengths (A) of these interactions, as well as a third hydrogen bond to a water molecule in the crystal structure are depicted by dotted lines. (Mathews, 1980). Proton N M R studies have also found a significant bias of conformation for protoporphyrin-containing 65 which is absent in 65 reconstituted with deuterohemin or isopemptohemin. These results suggest that the interactions between 21 Figure 6: Heme Orientational Disorder in Cytochrome D5 Two orientations of the heme prosthetic group are available in 6-type cytochromes owing to the absence of covalent bonds between the heme and protein. In the recombinant cytochrome 65 molecule, rotation about the a-7 heme axis results in a dominant orientation estimated at 60% versus 40% for the alternate orientation (Funk et a/.,1990b). the 4-vinyl group and the apo-protein may be responsible for the observed effect (LaMar et al., 1981). Studies of freshly reconstituted cytochrome 65 by Walker et al. (1988) have shown that the two conformers differ in reduction potential by 27 mV (+0.8 versus -26.2 mV vs. SHE) with the interconversion occurring with a half life of 12 hr. 22 The same heme reorientation takes almost 100 times longer in the NADH-reduced form of the protein (LaMar et al., 1978), thus the physiological relevance of this finding may be minimal. The lower third of the heme-binding fragment constitutes a separate hydrophobic core domain that has been highly conserved in all 65 species. Although no clear functional role has been described for this domain, evidence from N M R studies of apocytochrome 65 suggests that a stable native core centered about Trp22 exists in the absence of heme (Moore and Lecomte,1990). The hinge region and membrane anchoring domain have not been defined structurally, though Chou-Fasman analysis of the C -terminal primary sequence predicts the linker (91-97) to be a-helical (Fleming et al., 1978), while hydropathy plotting and proteolytic protection studies strongly suggest that residues 103-127 insert into the membrane, with residues 127-133 re-emerging into the cytosol (Ozols,1989) X-ray diffraction analysis of the dithionite-reduced protein (ferrocytochrome £5) showed virtually no difference in the positions of main chain and side chain atoms (Argos and Mathews, 1975). The only significant change in electron density occurs near the carboxylate of the buried propionate. This density was interpreted as indicating the binding of a cation upon reduction. The authors provided a model for this process in which the formal charge of +1 for the oxidized heme is neutralized by Coulombic interaction with the buried propionate, whereas upon reduction the heme loses charge and the heme propionate is free to attract a cation from solution (see Figure 7). In concordance with this model, the reduction potential of 65 is sensitive to ionic strength, p H and temperature (Reid et al., 1982), and elimination of the heme carboxylate charge 23 Figure 7: A Model for the Reduction/Oxidation Equilibrium of Cytochrome D5 (from Argos and Mathews, 1975). The reduction of cytochrome 65 by dithionite does not result in an appreciable rearrangement of the protein structure. This model suggests that the formal positive charge of the ferric atom in the oxidized species is stabilized by the carboxylate charge of heme propionate 7, which in turn is favoured by hydrogen bonds to Ser64 (see Figure 5). Upon reduction, the carboxylate charge is released from the iron center and is free to bind a cation from solution. Thus disruption of a hydrogen bond from Ser64 would be predicted to destabilized the carboxylate charge and the oxidized protein, shifting the equilibrium towards the, reduced cytochrome and raising the reduction potential. 24 H O O C C O O H Figure 8: Structure of Ferriprotoporphyrin IX Heme: R=H Dimethylester (DME) heme: R=CH3 25 by substitution of dimethylesterprotoporphyrin IX for protoheme IX in 65 increases the reduction potential by 60 mV (Reid et al., 1984). The physical properties of soluble cytochrome b$ (low molecular weight, high solubility and stability) make it ideal for N M R studies (Keller and Wuthrich,1980; Reid et al., 1987; Burch et al, 1990). The complete assignment of proton resonances of trypsin-solubilized hepatic cytochrome £5 from pig and calf liver has now been reported (Veitch et al., 1988; Guiles et al., 1990), and the resulting connectivity assignments are in excellent agreement with the crystallographic model reported by Mathews (1972). The anticipated availability of a solution structure of cytochrome 65 together with assignments for cytochrome c (Wand et al., 1989) should prove a powerful tool for investigating protein-protein interactions and charge transfer. 4. Nucleotide Sequence Analysis The elucidation of the nucleotide sequences of cDNAs encoding hepatic cytochrome 65 species has recently been achieved. Initial reports of the sequence for rabbit (Daruish et al, 1988) and chicken (Zhang and Somerville,1988) cDNAs demonstrated that the high degree of identity between species was, not surprisingly, conserved at the nucleotide level. The rabbit and chicken cDNAs exhibit a 76% nucleotide identity. The chicken cDNA sequence showed a suprising difference in that 5 additional amino acids were predicted at the N-terminus, while the C-terminus was one residue shorter than all mammalian 65 species. Thus, the chicken protein is estimated to consist of 138 residues, compared to 134 residues for all mammalian species. However the N-terminal sequence identified from chemical analysis of the protein (Ozols,1989) starts at the third residue predicted from the cDNA sequence. Evidently, a 26 proteolytic event removes the first two residues to yield a mature protein of 136 amino acids. During the course of studies reported in this dissertation, Steggles and co-workers used the rabbit cDNA as a cross-hybridization probe to isolate the corresponding hepatic cDNAs from bovine (Cristiano and Steggles, 1989) and human sources (Yoo and Steggles, 1988). These studies have correctly identified the N-terminal sequence of bovine 65, clarified the amidation status of several residues in this protein, provided the sequence of the hydrophobic region of the human protein and corrected the assignment of several residues. Southern analysis using the full-length chicken cDNA as a hybridization probe indicates the presence of multiple bands, while a probe spanning only the hydrophobic domain hybridizes to unique bands (Zhang and Somerville,1988). The authors suggest these results can be explained by the presence of two cytochrome 65 genes, only one of which encodes a hydrophobic domain. Alternatively, the hybridization pattern could result from a single gene in which exons encoding the polar and non-polar domains are separated by a 5.3 kbp intron. The characterization of several partially processed mRNAs for human b$ (Yoo and Steggles, 1989) shows that a splice junction occurs between residues 86 and 87. This observation is consistent with occurrence of a gene in which functional domains are commonly separated by introns (Gilbert, 1978; Rogers, 1985). Processed pseudogenes for human b$ have also been identified (Yoo and Steggles, 1989) and will no doubt complicate the search for active members of the 65 gene family. The presence of a distinct transcript encoding the erythrocytic form of cytochrome 65 has now been confirmed by sequence analysis of an erythrocytic cDNA (A. Steggles, personal communication). 27 B. The Transferrins The transferrins constitute a family of related iron binding proteins which share a common overall molecular structure, consisting of a two-lobed, single-chain molecule in which each half of the protein maintains a similar metal-binding site. The three principal members of the transferrin family of proteins include serum transferrin, which functions as the primary carrier of iron in the circulation (Shade and Caroline, 1946), lactoferrin, a major protein constituent of mammalian milk, tears and other secretions and ovotransferrin, one of the most abundant proteins in avian egg whites. The latter two species presumably function as anti-microbial agents by limiting the supply of free iron (Alderton et al.,1946). The two iron binding sites display high similarity in both structure and function. When compared at the level of amino acid and nucleotide sequence, the two lobes show high identity with each other, suggesting that an ancestral gene encoding a protein with a single metal binding site underwent a tandem duplication event to give rise to the present two lobed molecule (MacGillivray et a/,1983). Two-sited transferrin-related proteins have been found in all chordate species examined to date. The organization and chromosomal localization of the genes encoding transferrin family members strongly suggests that an ancestral gene encoding a two site transferrin molecule has undergone successive gene amplification events followed by divergence of both sequence and function to generate the present suite of transferrin-related genes (Bowman et al., 1988). This chapter describes in greater detail the structural and functional properties of the transferrins with particular emphasis on the molecular details made available during the past 10 years from studies on transferrin-encoding gene structures and the solution of the crystal structures for both the human lactoferrin and rabbit serum transferrin molecules. Throughout the remainder of this dissertation, the following abbreviations will be used: 28 TF: serum transferrin L F : lactoferrin OTF: ovotransferrin melanoTF: melanotransferrin, also p97 and melanoma-related transferrin T F / 2 N T F / 2 C : N and C lobe half molecules of transferrin Small case letters prefixing any of the above abbreviations will designate the species of origin (i.e. hTF: human serum transferrin). 1. Historical The presence of non-heme iron in plasma was described as early as 1925 (Fontes and Thivolle, 1925). The globulin fraction of plasma was later shown to be the principal carrier of iron (Starkenstein and Harvalik,1933), though the direct demonstration of a specific iron-binding protein was accomplished only after the development of a precise method for fractionating plasma constituents (Cohn et al., 1946). Independent studies by Holmberg and Laurell (1945) and Schade and Caroline (1946) were the first to demonstrate directly the capacity of a specific /3-globulin protein to bind ferric iron, which led to the application of some early synonyms for transferrin; siderophilin (Schade et al., 1949) and /91-metal-binding globulin (Surgenor et al., 1949). Holmberg and Laurell (1947) distinguished an iron-specific binding protein separate from the copper-binding component of plasma and were the first to suggest the name transferrin as such, a term which has superseded the earlier descriptors. Transferrins have been found in all chordate species examined to date. Typically, chordates maintain a similar suite of plasma proteins and even primitive members of these phyla, such as the nurse shark and hagfish (MacGillivray,1977, Aisen et al., 1972) , display a two-sited transferrin protein of molecular weight c.a. 80 kDa. A n examination of the prochordate sea squirt, Pyura haustor, by Yang et al. (1985) 29 demonstrated that TF-related sequences are found in the genome of this organism. Specific hybridization signals were identified on Southern blots of genomic D N A from the sea squirt, using either the N - or C-terminal coding regions of the hTF cDNA as a probe. These results may correlate with the presence of a "half-sized" TF-l ike iron binding protein (40 kDa) described in the related sea squirt Pyura stolonifera (Martin, et a/., 1984) and may represent the single-lobed iron binding protein postulated as the predecessor of modern day transferrins. Since the divergence of human and prochordate lineages is estimated at 400-500 million years b.p. (Doolittle,1984), transferrin-like sequences appear to have been remarkably well conserved throughout evolution. A n iron-binding component of avian egg whites which inhibited microbial growth was first described by Schade and Caroline (1946), and later identified as conalbumin (Alderton et al., 1946). Williams (1962) later recognized the similarities between conalbumin and chicken serum T F as the result of comparative studies employing gel electrophoresis, amino acid and carbohydrate composition analysis and immunoelectrophoresis. The term ovotransferrin (OTF) was suggested to reflect these similarities. More recently, the complete identity of the primary sequences of O T F and chicken serum T F has been confirmed by studies showing only a single T F gene within the genome (Lee et al., 1978), the only recognized difference between the two species being the carbohydrate composition (Williams, 1962). The presence of an iron-binding factor in the milk of certain mammalian species was confirmed by the isolation of a salmon-pink coloured protein from bovine milk (Polis and Shmukler, 1953) and from human milk (Johansson, 1958). This protein was shown to be immunologically distinct from serum T F (Johansson, 1960; Montreuil et.al., 1960) though it displayed a similar absorption spectrum and mode of iron binding. Schultze and Heremans (1960) encouraged the use of the term lactoferrin to describe the protein, reflecting the lack of antigenic similarities with serum T F , and thus replaced the terms lactotransferrin, lactosiderophilin and the "red protein". However the recent 30 solution of the crystal structure for human L F (Anderson et al., 1987) and rabbit serum T F (Bailey et al., 1988) has shown that the two molecules share a high degree of structural similarity, especially at the iron binding sites while analysis of the amino acid sequence of human lactoferrin (Metz-Boutigue et al., 1984) and the partial nucleotide sequence of the corresponding cDNA (Rado et al., 1987) shows a high degree of sequence conservation between these two proteins. In addition to lactating mammary gland, many other tissues express high levels of L F . Immunological detection has shown L F to be present in tears, bronchial secretions, seminal fluid and cervical mucus (Masson and Heremans,1966) where it likely serves as a bacteriostatic agent (Arnold et al.,1977). Lactoferrin has also been detected in the neutrophil lineage of human and guinea pig leukocytes (Masson et al.,1969) where it may serve a similar role in the acidic environment of these cells. A recent addition to the transferrin family of proteins is p97, initially identified as a cell surface glycoprotein which is over-expressed in human melanomas (Woodbury et al., 1980; Brown et al., 1980). p97 displays a protease-sensitive domain arrangement similar to other transferrins and also binds iron (Brown et al., 1982) but is not secreted; rather p97 exists as an integral membrane protein (Brown et al., 1981). The predicted amino acid sequence, deduced from the complete cDNA sequence (Rose et al., 1986) shows a typical signal sequence followed by two internally homologous domains, typical of other transferrins. A stretch of hydrophobic residues at the carboxy-terminus of the protein likely comprises a membrane-anchoring domain. A cellular function for p97 has not yet been identified as it is difficult to reconcile an iron transport role for an integral membrane protein, however the limited tissue distribution of the protein (melanomas, nevi and fetal intestine) suggests p97 may serve as a useful marker for these tissues. 31 2. Physiological Role and Expression Iron is of central importance to the chemistry of life. The ubiquitous requirement for iron is seen in reactions which include respiratory electron flow, the fixation of nitrogen and oxygen and ligand binding by proteins such as hemoglobin and myoglobin. The reversible reduction of ferric iron is often utilized by these systems to effect energy conservation and transfer, though under physiological conditions, F e m is the stable state. As noted by Aisen and Liebman (1973), the low solubility product of Fe(OH)3 (10~36 M) dictates that free iron in solution cannot exceed 10"1 7 M . As such, organisms have adapted several siderophilic molecules to perform the function of maintaining a high concentration of iron in soluble form; in chordates transferrins are the principal iron-binding proteins. In addition, the chelation of iron prevents iron-intoxication and minimizes urinary loss of iron (Putnam, 1975). The primary function of serum transferrin is to transport ferric iron from acquisition and storage organs to tissues where it is required. In human serum, transferrin binds almost all available iron, though this amount (3-4 mg) represents less than 0.1% of the body iron stores and leaves serum T F only thirty percent saturated (Gitlow and Beyers, 1952). The concentration of T F in serum is traditionally represented as a total iron-binding capacity (TIBC) and is calculated as the number of micrograms of iron bound by 100 mL of serum (Laurell,1960). The most studied model of iron transport remains the transferrin-reticulocyte system. Hemoglobin iron turnover accounts for almost 30 mg per day and reticulocytes express an estimated 300,000 T F receptors per cell in order to accommodate demand (van Bockxmeer and Morgan, 1979). The T F receptor molecule is well characterized, identified first by classical membrane biochemistry (see Newman et al., 1982) and later sequenced by characterization of genomic and cDNA clones (Kuhn et al.,1984; McClelland et al., 1984). The receptor consists of a homodimer of two disulfide-linked monomers of molecular weight 95 kDa each. Diferric T F binds with an apparent 32 affinity of 1 0 9 M _ 1 , with apoTF showing a 30-fold weaker affinity (Young and Aisen,1981). The complex is clustered into clathrin-coated pits (Watts,1985), likely as a result of interactions between the cytoplasmic tail of the receptor with coat proteins (Iacopetta et al., 1988). Invagination of these pits forms vesicles which then fuse with endosomes where conditions of low pH (5.5) cause the release of iron from T F (see Kuhn et a/., 1990). Transferrin is a true carrier molecule in that T F : T F receptor complexes are recycled to the cell surface (Harding et al., 1983; Hopkins and Trowbridge, 1983). Studies on several cultured cell types indicates that the limiting factor in cellular iron uptake is the number of T F receptor molecules on the membrane surface (Ciehanover et al.,1983; Iacopetta and Morgan,1983; Hebbert and Morgan,1985). Regulation is achieved by modulating the receptor mRNA activity through so-called iron responsive elements (IRE) in the 3' untranslated region of the mRNA ( Casey et al., 1988). IREs are palindromic sequences which bind a regulatory factor(s), leading to a stabilization of the mRNA molecule (Mullner et al., 1989; Rouault et al., 1988). This binding activity is iron-sensitive and thus links the translation of T F receptor m R N A to the level of available iron (Kuhn et al., 1990). In contrast, T F expression is only slightly sensitive to iron levels, showing a 2-fold increase in mRNA as a result of increased gene transcription (McKnight et al., 1980a; 1980b). However, O T F gene regulation by steroid hormones is well documented in the chick oviduct, where estrogen has been shown to effect an 8-fold rise in T F m R N A levels by increasing the rate of T F gene transcription in the primary oviduct, though the same treatment increases T F synthesis in the liver by only 2-fold, suggesting a tissue-specific component must also be involved (Oka and Schimke,1969; McKnight and Palmiter,1979). In the Sertoli cells of rat testis, follicle stimulating hormone, insulin, retinol and testosterone are all capable of increasing T F m R N A levels. Homology searches of the upstream region of the T F gene have identified several 33 potential hormone response elements (Lucero et al., 1986; Adrian et al., 1986) though their status as regulators of gene activity is still unclear. The liver is the primary site of serum protein synthesis, including transferrin (Morgan, 1983). The half life of the protein is approximately 8 days, suggesting a daily turnover of one gram (Jarnum and Lassen, 1961). Recent studies have used cell culture model systems in attempting to identify factors responsible for the liver-specificity of T F synthesis. Promoter analysis of the T F gene has shown the region from -650 to the transcriptional start site as essential for hepatocyte specific activity and several protein-binding domains close to the start site have been demonstrated by footprint analysis (Brunei et al., 1988). In addition, T F is highly expressed in the Sertoli cells of the testis, the fetal yolk sac and the choroid plexus, all examples in which T F is synthesized in tissues adjacent to physiologic enclosures ( i.e. the testes ducts, placenta and cerebrospinal system, respectively) in order to maintain homeostasis within the compartment (see Funk and MacGillivray,1990; Bowman et al., 1988). T F expression has also been demonstrated in T-lymphocytes, submaxillary gland, ovary, spleen, bone marrow, lymph nodes and thymus (Lum et al., 1986; Thorbecke et al., 1973; Phillips and Thorbecke,1966; Meek and Adamson,1985) Transferrin has also been demonstrated to act as a potent antimicrobial agent (Schade and Caroline, 1946) though L F is a much more likely candidate for this role, owing to its wide distribution in body fluids, tighter iron binding and greater acid stability (Mason et al., 1965,1966). In concordance with this role, lactoferrin has been identified in many biological fluids including tears, bronchial secretions, seminal fluids and cervical mucus, and thus represents a major secreted protein of endothelial cells (Masson and Heremans,1966; Arnold et al., 1977). 34 3. Fundamental Properties i) Physiochemical and Conformational Characteristics The relatively high concentration of transferrin in serum and the central role played by the protein in iron transport led to intensive studies on the isolation and characterization of serum T F soon after the initial description by Holmberg and Laurell (1945). Serum T F belongs to the /9-globulin fraction, having a typical serum concentration of 250 mg/lOOmL. Some of the earliest purification methods involved crystallization of T F directly from ethanol fractions of serum (Surgeoner et al., 1949; Koechlin, 1952). Later, methods were developed for large-scale purification, involving the use of ammonium sulfate, ethanol or 2-ethoxy-6,9-diaminoacridine lactate (Rivanol) precipitations, ion exchange and gel exclusion chromatography or combinations of these techniques (see for example Inman et al.,1961; Schultze and Schwick, 1957). Serum T F was readily identified by electrophoretic methods, such as the discontinuous buffer starch gel system described by Poulik (1957) or in combination with radioautography using 5 9 F e (Giblett et al., 1959). However, estimates of the molecular weight of serum transferrin differed widely during this period. Sedimentation diffusion and sedimentation equilibrium methods were often employed, resulting in initial estimates between 68-90 kDa for human serum T F (Schultze and Schwick, 1957; Charlwood, 1963). Other methods, such as gel filtration, intrinsic viscosity, iron-binding capacity and NaDodS04 gel electrophoresis arrived at values within this range (see Putnam, 1975), though it was not until more accurate sedimentation equilibrium measurements became available that a value of approximately 77 kDa was generally accepted (Mann et al., 1970; Aisen et al., 1970). Other accepted physical properties of serum T F include: a partial specific volume of 0.725 (Oncley et al., 1947), a diffusion coefficient of 5.4 Fick units (Bezkorovainy and Grohlich, 1967), and a frictional ratio (f/fo) of 1.37 (Oncley et al., 1947). The latter determination suggests an asymmetric shape for the molecule, in agreement with later studies which determined the protein 35 shape as ellipsoidal, with a major to minor axis ratio of 2:1 (Rosseneu-Moutreff et al., 1971). Evidence from many studies show a significant conformational change in transferrin molecules upon binding of iron. Ferric complexes of T F and O T F are more resistant to denaturation and have reduced susceptibility to proteases (Azari and Feeney 1958; 1961; Yeh et al.,\919) and hydrogen-tritium exchange (Ulmer,1969). The major to minor axis ratio increases from 2.0:1 to 2.5:1 (Rosseneu-Montreff et al., 1971) which confirms recent crystallographic evidence showing an open configuration for the metal free N lobe of apoLF (Baker et.al.,\9Z9). As well, saturation with iron lowers the isoelectric point of T F from 5.80 to 5.45 (Keller and Pennell,1959). ii) Carbohydrate Composition The carbohydrate content of human serum T F was initially reported by Heimburger et al. (1964) and later studies involving neuraminidase treatment demonstrated the presence of terminal sialic acid residues (Blumberg and Warren, 1961; Poulik, 1961). The complete sequence of the oligosaccharide was later reported by Jamieson et al. (1971) though later investigations involving chemical and enzymatic studies (Spik et al., 1975) and nuclear resonance spectroscopy (Dorland et al., 1977) and mass spectroscopy (Karlsson et al., 1978) have identified a different biantennary structure (Figure 9). The number of heteropolysaccharide chains varies between transferrins of different species (Graham and Williams, 1975). Human serum T F contains two identical chains, linked to asparagine residues 415 and 608 in the C lobe of the protein (MacGillivray et al., 1977). As yet no functional role has been ascribed to the carbohydrate moieties of T F . Enzymic removal of terminal sialic residues with neuraminidase does not result in gross changes in conformation (Bezkorovainy,1966), nor does it interfere with iron binding 36 NeuNaca(2-^)Galp(H4)GlcNAcp(l->2)Mana(l-*3) >^ Manp(B4)GlcNAcp(H4)GlcNAcpl->Asn NeuNaca(2^)Galp(H4)GlcNAcp(l->2)Mana(l->6) Figure 9: Structure of the Glycan Chains of Human and Rabbit Serum Transferrin. As determined by Leger et al. (1978), the predominant carbohydrate structure of serum transferrin is doubly branched (biantennary), with each branch terminating with a sialic acid residue. The two carbohydrate chains of human serum T F are both attached to the C lobe via asparagyl linkages to residues 415 and 608. (Morgan et al., 1967). Transferrin differs from most other plasma glycoproteins in that neuraminidase treatment does not enhance clearance by the liver (Ashwell and Morell,1974). It has been suggested that oligosaccharide may function to enhance the secretion of proteins to which they are attached, however treatment of rat hepatocytes with tunicamycin causes only a slight decrease in T F secretion even though only 25% of T F is glycosylated under these conditions (Struck et al., 1978). Attempts to demonstrate a role for the carbohydrate moiety in the binding of T F to its receptor have not been successful. iii) Amino Acid Sequence Analysis The amino acid composition of human serum T F has been the subject of many investigations (Parker and Beam, 1962; Roop and Putnam, 1967; Palmour and Sutton, 1971; Sutton and Brew, 1974). These studies indicated the protein consisted of approximately 630 amino acid residues, with a calculated molecular weight of 76 kDa. Other significant findings from composition analysis studies include a complete absence of free sulfhydryls and a substantially greater proportion of basic amino acids in L F as 37 compared to serum T F , a fact reflected in the higher pi of L F (pl=5.3 versus 5.8 for human serum TF) (Putnam, 1975). Amino-terminal sequence analysis from several species provided clear evidence of the high degree of sequence conservation in the transferrin family (see Graham and Williams, 1975; Jolles et al.,\916). Partial sequences for tryptic and cyanogen bromide fragments from human T F , O T F and L F were complicated by the presence of multiple half-cystine residues (see Putnam, 1975), however MacGillivray et.al. (1977) were able to show conclusively the presence of a high degree of internal sequence identity between the two lobes of human serum T F . The same group later reported the complete primary sequence of the protein which indicated a remarkable 40% identity between residues of the N and C lobes (MacGillivray et dl., 1983), providing indisputable evidence for at least one gene duplication event in the evolutionary past of T F . The complete primary sequence of human L F (Metz-Boutigue et al., 1984) similarly showed a high degree of conservation (37% between the two lobes) and an even higher degree of identity between the complete human serum T F and L F sequences (59%) (Figure 10). iv) Metal Binding Characteristics The binding of two atoms of ferric iron by each T F molecule involves the concomitant binding of two carbonate anions (Schade et al., 1949) resulting in a salmon-pink complex characterized by an absorption maximum at 465 nm (Schade and Caroline, 1944,1946; Cohn,1947). As the precise nature of the iron binding capacity of melanotransferrin has yet to be resolved, the following discussion will be restricted to serum T F , O T F and L F . The available crystal structures for human L F (Anderson et a/., 1987; 1989) and rabbit serum T F (Bailey et al., 1988) have clearly identified the protein ligands to the iron (consisting of two tyrosyl, one histidyl and one aspartyl residue) and have confirmed a bidentate carbonate anion as the bound anion (see IB6. Iron Binding Site and 38 hSTF rSTF hLF cOTF hTF 41 -- -- --rTF* 81 N.D. -- - -hLF* 59 57 37 --cOTF 51 50 49 33 * 3D-crystallographic structures for human lactoferrin (Anderson et al.,1987) and rabbit serum transferrin (Bailey et al.,1988) have been reported. Figure 10: Amino Acid Sequence Identity Between Transferrin Family Members. Primary sequence data for the mature forms of human serum transferrin (hTF) (Yang et al.,1984), human lactoferrin (hLF) (Metz-Boutigue et al.,1984), chicken ovotransferrin (cOTF) (Jeltsch and Chambon,1981) and the first 397 amino acids of mature rabbit serum transferrin (rTF) (D.K. Banfield, unpublished results) were compared using the method of Meyers and Miller (1988) and are reported as percentage amino acid identities. Comparisons of the N and C lobes of individual species appear on the diagonal. Structural Considerations). Previous physiochemical studies however have implicated additional residues as being involved in the iron binding process, several of which are described below. The iron:TF complex is sensitive to pH, being stable in the range from 7.5 to 10, but unstable below 6.5 (Ehrenberg and Laurell,1955). The binding affinity is extremely tight; under physiological conditions the K a of the N lobe binding site for ferric iron estimated at 10 2 2 M" 1 with the C lobe site having approximately five-fold greater affinity (Aisen and Leibman,1978). Indeed this binding is so strong that iron bound at either site is effectively locked, such that kinetic, rather than thermodynamic considerations govern the relative occupancy of the two sites (Aisen and Brown, 1975a). The binding of iron to lactoferrin has been estimated to be 260-fold stronger than serum T F , and 39 displays considerable stability at acidic pH (2-4.5) (Aisen and Leibman,1972), being fully dissociated only at pH 2 (Johansson, 1960). Transferrin also binds a variety of transition and lanthanide series metals, including C u + 2 , Z n + 2 (Conn, 1947), C r + 3 , M n + 2 , C o + 2 , N i + 2 , C d + 2 , (Tan and Woodworth,1969; Aisen et al, 1969), G a + 3 (Woodworth et al, 1970), T b + 3 , E u + 3 , E r + 3 , H o + 3 , N d + 3 , P r + 3 (Luk, 1971). The relative affinity of several of these metals was determined by Tan and Woodworth (1969) as: F e + 3 C r + 3 > C u + 2 M n + 2 > C o + 2 > C d + 2 Z n + 2 N i + 2 Complexes with chromium, manganese and cobalt show binding of a molar equivalent of carbonate. Though the ferric complex is likely the only one of physiological significance, reports of T F in serum binding chromium (Hopkins and Schwartz, 1964), zinc (Harris, 1983) and manganese (Aisen et al.,1916) have been reported. Ferrous iron is bound very weakly to T F , if at all (Gaber and Aisen, 1970). Prior to the resolution of detailed crystal structures, much emphasis in T F research was placed on the identification of the binding site ligands. These studies used many different techniques, including N M R spectroscopy, resonance-enhanced Raman spectroscopy, proton titration, EPR spectroscopy and several chemical modification approaches (see Aisen and Brown, 1975b; Chasteen, 1977). Candidate ligands included up to three phenolic groups (Warner and Weber, 1953), two histidine imidazole groups (Aasa et a/.,1963) and possibly a tryptophan indole ring (Tomimsatsu et al., 1973). Though the identity of the binding ligands is now conclusive, residues removed or even distant from the metal may yet prove to be critical for normal iron binding. For example a Gly65Arg variant of serum transferrin has a characteristic yellow absorption maximum, likely due to an alteration of the electronic environment near the Asp63 ligand (Evans and Madden, 1984). The third tyrosine residue implicated in several earlier studies may prove to be Tyr83 (hLF numbering system), as this residue is invariant in both lobes of all T F species and is situated within 6A of the metal (Baker et al, 1987) Residues 40 involved in fixing the anion would also be crucial for normal iron binding, as evidenced by chemical modification studies in which an essential arginyl residue participates at the metal binding site (Rogers et al.,\911), a feature confirmed by crystal structure determinations. Nuclear magnetic resonance spectroscopic studies on whole molecule transferrin species (Markley,1975; Alsaadi et al., 1987; Woodworth, 1986) and half molecules (Valcour and Woodworth, 1987) have shown histidyl residues to be directly affected by iron binding. The presence of a suitable anion is an absolute requirement for specific iron binding (Bates and Schlabach,1975). The binding of metal and anion is cooperative, as neither is bound in the absence of the other (Price and Gibbons, 1972; Aisen et al., 1973), though many small anions can be substituted for carbonate in the complex. These include oxalate, malonate, E D T A , glycinate, thioglycolate and nitrilotriacetate (Aisen et a/.,1967; Woodworth et al., 1975, Schlabach and Bates,1975). These are all molecules that share the structural characteristic of having at least one carboxylate function and a second functional group in close proximity. These functionalities likely act as metal ligands, reflected in the finding that the ligand fields and iron-binding affinities vary with the bound ligand (Pinkowitz and Aisen, 1972). In addition, the synergistic anion also functions to shield the ferric ion from local positive charge within the binding site. As shown in the h L F crystal structure (Anderson et al., 1987), the carbonate dianion is in close proximity to the invariant Argl24 side chain and additional positive charge is provided by a dipole effect from the N-terminus of nearby a-helix 121-137. N M R studies have also shown anion binding likely precedes metal binding (Woodworth, 1975). Proteolytic digestion of several species of serum transferrin and lactoferrin releases a small (20 kDa) iron-binding fragment which includes domain II of the protein as well as portions of the bridging strands (Evans and Madden, 1984; Legrand et al., 1984). These polypeptides do not contain the Asp ligand provided by domain I, and display yellow-shifted absorption maxima and reduced iron-binding affinities. The 41 presence of a synergistic anion is required for iron binding and binding to these fragments is specific and reversible (Legrand et al., 1990). A plausible model for iron binding suggests that the initial complex includes the anion, metal ion and ligands contributed from domain II and the bridging strands. Hinge-like movement of the two domains closes the lobe as the Asp ligand coordinates, thereby locking the metal in a tight complex (Baker et al., 1987). Several mechanisms may effect iron release, including protonation of the side-chain ligands, direct attack on the anion, chelation via ligand exchange and reduction of the ferric ion (Aisen and Listowsky,1980) v) The Two Lobes of Transferrin The preparation of 35 kDa iron-binding fragments from O T F by limited proteolysis demonstrated conclusively the presence of two structurally distinct lobes in transferrin molecules (Williams 1974,1975). This finding was later confirmed by low-resolution x-ray diffraction data which showed two lobes of similar size and structure, best approximated by two ellipsoids of revolution inclined at 30° to each other (Gorinsky et al.,\919). The independence of metal binding by the two lobes was demonstrated conclusively by the introduction of the urea-polyacrylamide electrophoresis procedure of Makey and Seal (1976) which allowed the resolution of mono-ferric forms of transferrin molecules which have slightly different electrophoretic mobilities. A large number of studies have sought to determine whether the two lobes of T F differ in functional properties. A n initial impetus for this work was the reported finding of Warner and Weber (1953) of strong cooperativity between the two metal-binding sites (K2=100Kj), though this finding was later refuted by the demonstration of distinct mono-ferric forms of the molecule (Aisen and Listowsky,1980). However many investigators have now reported slight but real differences between the N and C lobes (Brock, 1985a). EPR studies of Cr+ 3 - (Aisen et al, 1969) and VO + 2 - loaded T F (Cannon 42 and Chasteen,1975) showed clear differences between the two sites as have Mossbauer and optical spectroscopic measurements (Aisen et al., 1973). The pH-dependence of iron binding also differs. The C lobe is more susceptible to proton attack with a five-fold lower affinity at pH 7.4, but a 33-fold lower affinity at pH 6.7 (Lestas,1976; Princiotto and Zapolski,1975). In addition, the two lobes differ in their ability to bind large ionic radii metals such as N d + 3 and P r + 3 (Luk,1971) and various anions, such as oxalate (Ainscough et al., 1983). Structural studies have shown differences between the two lobes near the distal regions of the iron-binding clefts which may affect the accessibility of the two sites for small molecules, while the presence of an additional disulfide bond in the C lobe likely accounts for the greater stability and higher iron affinity of that lobe (Baker et al., 1987). The production of a "half-molecule" of transferrin, corresponding to the N lobe of the intact molecule was first achieved by Williams (1974), who capitalized on the finding that partially saturated O T F selectively loads the N lobe site, leaving the vacant C lobe in a loose conformation which is susceptible to proteolytic digestion. The resulting N lobe fragment (OTF/2N) maintains the ability to bind metals and was also reported to donate iron to reticulocytes. The isolation of the corresponding C lobe (OTF/2C) was achieved by tryptic digestion of iron-saturated O T F under mildly acidic conditions, whereupon the N lobe is selectively relieved of iron (Williams, 1975). The O T F / 2 C lobe was also shown to bind metals and donate iron to reticulocytes and also carries the carbohydrate moiety. Isolation of the corresponding half molecules of human serum T F (Evans and Williams, 1978; Lineback-Zins and Brew,1980; Zak and Aisen,1985) and bovine serum T F (Brock et al., 1978) followed; however, individual bovine half molecule species were incapable of donating iron to rabbit reticulocytes, whereas diferric bovine T F could donate iron. A similar result was later reported by Brown-Mason and Woodworth (1984) who, in contrast to previous studies, failed to observe donation of iron to reticulocytes by O T F half molecules, suggesting that each lobe of the O T F 43 molecule contains a distinct recognition region necessary for receptor binding and subsequent iron donation. 4. Nucleotide Sequence Analysis Investigations carried out over the past ten years on the gene and c D N A sequences of transferrin family members has proven valuable in identifying mechanisms by which gene families evolve. Comparison of coding sequences, exon organization and chromosomal localization have provided finer resolution of the lineage of the transferrin gene family and have confirmed that both intragenic duplication and gene amplification events have occurred during the evolution of modern transferrins. The availability of D N A probes has also augmented studies of the expression patterns of transferrins, the results of which have been recently reviewed (Bowman et al., 1988, Funk and MacGillivray, 1990). The following section relates some of the important findings of studies on the molecular biology of transferrins, with particular emphasis on the evolution of this gene family. i) Chicken Ovotransferrin The chicken O T F gene was the first transferrin species investigated at the nucleotide level. Southern blot analysis of chicken genomic D N A demonstrated that serum T F and O T F are the product of a single gene consisting of at least 17 exons separated by at least 16 introns (Cochet et al.,\919). This observation has now been confirmed by the determination of the complete nucleotide sequence of the ovotransferrin gene (Jeltsch and Chambon,1979), and provides direct evidence that the internal homology between the amino- and carboxy-terminal domains arose as the result of the duplication of an ancestral iron binding protein, as had been suggested previously for human serum transferrin (MacGillivray and Brew,1975). The suggested scheme for this event is identical to that proposed for the evolution of the human serum 44 transferrin gene (Park et al.,1985), involving the intragenic duplication of exons 2-9 from the 9 exon ancestral gene followed by the gain of exon 11 in the second domain (see Figure 11). Several conserved histidyl and arginyl residues in human transferrin have been proposed to contribute directly to the iron binding sites, and their occurrence has been postulated to result from an even more ancient quadruplication event, such that a primordial iron binding structure originally duplicated to yield the first lobe of transferrin, then duplicated again to produce the present structure (MacGillivray et al., 1975). When the coding sequences for these fourfold homology blocks in the chicken ovotransferrin c D N A are compared, the sequence identity between any pairs varies from 37 to 67%, a value far higher than would be expected at random, considering that the proposed first duplication in this scheme would necessarily be dated well prior to 500 million years b.p. (Bowman et al., 1988) (see Figure 12). However these residues do not occupy "equivalent" positions in the domains of a given lobe (Baker et al.,1987) and thus this finding cannot be substantiated through structural studies. The promoter region of the ovotransferrin gene displays a typical eukaryotic T A T A box at position -31 (Cochet et al.,\919). Chicken ovotransferrin m R N A is 2376 nucleotides in size and contains a base-paired loop (nucleotides 2-29) of unknown function at the 5' end. This is followed by a short region complementary to 18S r R N A . A polyadenylation site is found at the distal end of the 182 bp 3'untranslated region. Comparison of the gene and cDNA sequences has revealed several sequence dissimilarities resulting in 3 predicted amino acid differences. These are the most likely causes of the known polymorphisms found in ovotransferrins (Williams et al., 1982). 45 1 2 3 4 5 PRIMORDIAL DUPLICATION EVENT ??? X ( » 5 0 0 m.y.) 1 2 3 4 5 6 7 8 9 10 1 A2 3 4 5 6 7 8 9 10 i INTRAGENIC DUPLICATION EVENT (>500 m.y.) 1 2 3' 4 5 6 7 8 9 2 3" 4 5 6 7 8 9 10 p97 LACTOFERRIN GAIN OF EXON 11 IN C-TERMINAL LOBE DIVERGENCE W \ CHICKEN OTF B-lym 3 4 5 6 7 10 11 12 13 14 15 16 17 PRESENT hTF GENE 46 Figure 11: Possible Scheme for the Evolution of the Present Human Serum Transferrin Gene (based on Park et al.,1985). Certainly one, and possibly two intragenic duplication events have occurred to give rise to present day transferrin genes. The presence of a "half-sized" transferrin-like protein in prochordates (Bowman et al., 1988) suggests the precursor iron-binding molecule was present prior to the divergence of this lineage from the vertebrates (500 m.y.). Sequence comparison of the two domains of serum transferrin provides weak evidence that this single lobe molecule may itself be the result of an even earlier duplication event (see Figure 12). Following the assemblage of a two-lobed transferrin gene, this progenitor sequence gained an additional exon within the C-terminal lobe, followed by several gene amplification events which gave rise to the modern suite of transferrin-related gene sequences. ii) Human Serum Transferrin and Lactoferrin The overall organization of the human gene is the same as that seen in the chicken gene (Park et al., 1985: Breathnach et al., 1978). When compared to their ovotransferrin counterparts, the 17 exons show very similar sizes, and the intron/exon splicing pattern is identical for the homologous sequences in the two domains of the human transferrin sequence, providing further evidence for the gene duplication event. The human gene spans 33.5 kbp. Virtually all of the size difference compared to the ovotransferrin gene is because the introns are larger in the human gene. The human transferrin m R N A is 2.3 kbp in size, including 54 bp of 5' untranslated sequence and 170bp of 3' untranslated sequence (Williams et al., 1982: Schaeffer et a/.,1987). The 5' flanking region of the human transferrin gene contains a near-canonical C A A T transcription factor (CTF) recognition sequence (Jones et al., 1987) at position -187 relative to the transcriptional start site. This region binds purified C T F protein (Brunei et al., 1988). Other consensus regulatory sequences have been identified in the 5' flanking region of the human transferrin gene, but their functional significance is unclear (Lucero et al., 1986). The cloning of a partial cDNA containing approximately 40% of the coding sequence for human lactoferrin has been reported (Rado et al., 1987), and it shows almost complete agreement with the previously reported protein sequence. I 116 HlsThrGlyLeuGlyArgSerAlaGlyTrpABnlleProIleGlvThrljeuLeu 133 II 250 Hl^AlaValyalAla^gAapAspAsiiLysValGluABpIleTrpSerPheLeu 267 III 455 HisThrAlaValGlyAraThrAlaGlyTrpVallleProMetGly Leulle 471 IV 592 HlsAlaValValvalArgProGluLysAlaAsnLys 11eAraAspLeuLeu 608 I 479 CMACGGGGCUGGGCATCUCUGCGGGCUGGAACAUCCCCATOGGGACACUCCSC 532 II 881 CACGCCGUCGUGGCUCGGGAUGACAACAAGGUUGAAGAUAUCUGGAGCUTJCCUC 934 III 1496 CACACCGCUGUGGGGAGGACtTGCUGGCUGGGUCAUCCCCAUGGGC UUGAUTJ 1546 IV 1907 CACGCUGUGOTCGUGCGCCCGGAGAAAGCAAACAAA AUCCGUGAUCUGCUG 1957 COMPARISON OF THE FOUR REGIONS l/ll l/lll l/IV ll/lll ll/IV lll/IV LENGTH IN NUCLEOTIDES 54/54 54/51 54/51 54/51 54/51 51/51 NUMBER OF CODON DELETIONS -/- •'1 -/1 -/1 -/I 1/1 NUMBER OF IDENTICAL 24 36 24 24 25 20 NUCLEOTIDES % NUCLEOTIDE IDENTITY 44 67 44 44 46 37 Figure 12: Sequence Comparison of the Four Domains of Ovotransferrin (from Jeltsch and Chambon, 1981) Panel A: The amino acid and nucleotide sequences of Domains I-IV of chicken O T F have been aligned to demonstrate the putative four-fold homology of transferrin molecules. Single codon gaps in Domains III and IV were introduced in order to maximize the homologies. Similar homologies have been described for human serum T F (MacGillivray et al., 1977). Panel B: Summary of the nucleotide sequence identities from Panel A . Nucleotides which occur in at least three of the Domains are scored as identical. The genomic arrangement of the transferrin gene and related gene species reflects their common evolutionary ancestry, as the human genes for transferrin, lactoferrin, p97 and the transferrin receptor are all found on the same chromosome. 48 Hybridization of a labeled human transferrin cDNA to metaphase chromosome spreads localized the human transferrin gene to the 3q21-25 region of chromosome 3 (Yang et a/., 1984). The long arm of chromosome 3 also contains the genes for the transferrin receptor (Goodfellow et al., 1982), p97 (Plowman et al., 1983) and lactoferrin (Teng et al, 1987) (see Funk and MacGillivray,1990). These results clearly indicate a common ancestry for members of the transferrin gene family. Presumably, the gene for a primordial iron-binding protein has been duplicated several times. Initially, a gene duplication gave rise to the present day transferrin molecule consisting of two homologous domains each with an iron-binding site. Subsequent gene duplications and divergence then gave rise to the other members of the transferrin gene family such as lactoferrin and p97 (see Figure 11). 5. Molecular Structure of Transferrin and Lactoferrin The recently published crystallographic structures of iron-saturated human lactoferrin (Anderson et al.,1987) and rabbit serum transferrin (Bailey et al.,1988) have corroborated and clarified many of the previous physiochemical and structural studies of transferrin molecules. As expected, the two available crystal structures clearly define two distinct iron-binding lobes in each molecule connected by a short helical linker sequence, and have allowed absolute determination of the iron ligands at each site. The initial description the h L F structure was subsequently used to aid in the interpretation of a 3.3A resolution map for the rabbit T F structure and though the complete primary sequence for the rabbit serum T F molecule is currently unavailable, it has been possible to extract a detailed structural analysis of serum T f as well. i) Human Lactoferrin The human lactoferrin crystallographic structure provided conclusive evidence that the high degree of primary sequence identity between the two halves of the coding 49 sequence is conserved in the overall three dimensional arrangement of the protein. Each lobe contains a single iron binding site with four amino acid side chains (Asp61, Tyr93, Tyrl91 and His252 in the N lobe; Asp407, Tyr447, Tyr540 and His609 in the C lobe) contributing ligands to the metal (Figure 13). Further refinement of the structure has conclusively identified a bidentate carbonate ion filling the remaining two coordination positions (Anderson et al., 1989). The two lobes of the molecule are arranged in an approximately antiparallel configuration, such that one lobe can be superimposed following a rotation of 180° and translation of 25A (see Figure 14 for the equivalent organization of rabbit serum TF). Each lobe in turn is composed of two structural domains (termed Nl.NII, CI and CII after the nomenclature proposed by Anderson et a/.,1987)(see Figure 14). In examining the structural organization of the N lobe (Figure 15), the alpha-carbon backbone begins in domain NI (residues 1-90) as a series of /J strands with connecting loops and helices, passes into domain N i l via a strand composed of residues 78-100, continues as domain N i l (residues 91-251), returns to domains NI via a second connecting strand (244-257) and completes the lobe in domain NI (residues 252-320) A final short helix (residues 320-331) runs across the lobe towards domain N i l and leads into the connecting helix which joins the lobes. Minor differences in the topology of the N and C lobes are restricted to the external loops and do not affect the overall folding pattern. The N lobe contains 6 disulfide bridges, and these are maintained in equivalent positions in the C lobe suggesting that this pattern arose prior to the gene duplication event that gave rise to modern transferrins. In addition the C lobe contains four additional cystine residues, two of which are involved in folding the terminal residues back over the surface of the C lobe. None of the disulfides span between the two lobes, nor do they bridge across the two domains except at or near the strands which connect domains. Carbohydrate attachment sites have been identified at Asnl37 and Asn490, though the density map does not allow for accurate structural definition of the glycan moiety. The overall 50 Figure 13: Metal Ligands of Transferrin The four side-chain ligands and carbonate dianion of the N-terminal lobe of rabbit serum transferrin are shown as ball-and-stick representations, with the surrounding peptide backbone represented as ribbon strand. Tyrl88 is contributed from Domain II, Tyr95 and His249 are from the bridging strands which connect the two domains, and Asp63 originates from Domain I. configuration of each lobe therefore is hinge-like, involving two domains which are restricted in conformation by multiple cystines and are connected by two flexible strands. ii) Rabbit Serum Transferrin The general organization of rabbit serum transferrin is the same as that of human lactoferrin. The alpha-C chain follows the same path as that described for h L F with Figure 14: Overall Molecular Organization of Rabbit Serum Transferrin The four domains of rabbit serum T F are demonstrated in this a-carbon representation. Domains NI (yellow) and N i l (pink) of the N-lobe are connected with Domains CI (green) and CII (orange) of the C-lobe by a linker peptide (dark blue). The iron atoms and accompanying bicarbonate anions are shown in red and sky blue respectively. The two lobes of the molecule are related across a screw axis of 167° and 25 A translation. domain II of the N lobe being composed of residues 97-245; domain NI is composed of residues which flank domain Ni l in the primary sequence. A total of nineteen cystine residues are found in serum T F , sixteen of which are located in the same relative position in hLF. Of the three additional disulfides, one is a short-range bridge found in the N lobe, while the other two are located at the ends of the interlobe connecting peptide. The presence of additional cystine residues in this linker peptide may confer conformational restrictions upon the serum T F molecule, which is reflected in the rotation required to superimpose the two lobes of the molecule (167° rotation, 23A 52 | hTF/2N molecule linker peptide— Figure 15: Structural Organization of the N-terminal Lobe of Rabbit Serum Transferrin In this ribbon representation of the N lobe of transferrin, the two domains (NI and Nil) of the molecule are shown to be composed of ^S-sheet structures joined by connecting loops and helices. In turn, the two domains are themselves connected by two bridging strands. In the full two-lobed molecule, the linker peptide joins the two lobes across the NI and CI domains. The binding of iron (red) involves a hinge-like movement of the two domains about the connecting strands thus trapping the metal deep within the protein interior. translation for serum T F versus 180° rotation and 25A translation for hLF). Whether these differences are of functional significance is as yet undetermined. Rabbit serum T F reportedly possesses either one or two carbohydrate moieties (Leger et al., 1978; Strickland et al., 1979). The crystal structure of the molecule can be interpreted in terms of at least one glycan structure at or near positions 491-495 in domain CII; however, the structure lacks precise definition. 53 The availability of gene sequences for both hen ovoTF (Cochet et al.,\919) and human serum T F ( Park et al., 1985; Schaeffer et al., 1987) has shown a high degree of conservation in the exon organization of T F genes from species distant in evolution. Though information on the organization of the rabbit serum T F gene is presently unavailable, an attempt has been made to correlate the exon organization of the human T F gene with the presence of discrete structural elements within the rabbit crystallographic model (Bailey et al., 1988) The N-terminal signal sequence is almost completely encoded by a single exon (Exon-1) while the remaining exons consist of seven homologous pairs common to both lobes, and two exons unique to the C lobe. Each amino acid ligand of the metal binding site is contributed by a separate exon, an unusual feature for eukaryotic metal binding proteins in which ligands are frequently clustered in discrete domains. Exon-2 of the N lobe (Exon-9 in the C lobe) displays high identity with single exons in both the chicken and human Blym oncogenes. Conserved residues in this region of the Blym transforming proteins are also those conserved between different transferrin species. This portion of the T F molecule forms a discrete supersecondary structural element which is remote from the iron binding site and therefore may be required for functions unrelated to metal binding (see Figure 16). Since the homologous regions in the transforming proteins constitute the majority of the primary sequence (56 of 65 residues), exon-2 may therefore retain a function in the T F molecule, such as interaction with a receptor, which is independent of iron binding. 6. Iron Binding Site and Structural Considerations The elucidation of the hLF crystal structure unequivocally identified the protein ligands involved in iron binding and in doing so clarified the interpretation of data from previous physiochemical studies. As the protein ligands in both the h L F and rabbit 54 Figure 16: Region of the N Lobe Encoded by Exon 2. Exon 2 of the human transferrin gene encodes a distinct super-secondary structure (yellow) in the mature molecule which is remote from the iron-binding site (red) (Bailey et al., 1988). This region shows high identity with both the human and chicken Blym transforming proteins (Diamond et al., 1984; Goubin et al., 1983). In lactoferrin, Domain NI of the N-terminal lobe is the primary recognition site for the lymphocytic lactoferrin receptor (Rochard et al., 1989), raising the possibility that this interaction may involve the exon 2 domain. serum T F molecules are identical and occur in equivalent positions in the two molecules, the following discussion applies equally to both. In the N lobe of the hLF molecule, Asp61, Tyr93, Tyrl91 and His252 provide ligands to the iron through their respective side-chain functionalities (equivalent positions in the human and rabbit serum T F molecules are Asp63, Tyr95, Tyrl88 and His249) while the C lobe ligands are contributed by the equivalent residues in that half of the molecule. Asp61 is contributed by a loop in Domain NI. The presence of a 55 carboxylate ligand was not anticipated from previous chemical studies. The replacement of this residue with serine in the equivalent position in the C-terminal lobe of melanoTF suggests this lobe of the molecule may not bind iron or would bind iron in a manner different to other transferrin lobes. Domain Nil provides the Tyrl91 ligand from the N-terminus of a helical segment and the invariant Argl21 residue from the N-terminus of another helical segment. The remaining two ligands (Tyr93 and His252) originate from the two domain-connecting strands. The final two coordination positions of the iron atom are occupied by a bidentate carbonate ion in the refined version of the hLF structure. The counterion occupies the coordination positions closest to the opening of the cleft between the two domains, and as such is considered to bind the "open" side of the iron atom. This also places the carbonate atom in a position to interact directly with the invariant Argl21 as had been anticipated from modification studies (Rogers et al., 1978) and may also interact with the dipole moment at the N-terminus of an adjacent helix. Several other amino acid side chains identified within the iron-binding cleft have been implicated as being necessary for iron and counterion binding. These include the invariant Tyr83 residue, likely the "third Tyr ligand" identified in chemical studies (Aisen and Listowsky,1980; Brock,1985b; Chasteen,1983), and Gly63, which has been substituted by Arg in the C lobe of a naturally occurring variant of human serum TF whose visible absorption spectrum is significantly shifted (Evans et al., 1982). As well, Lys206 of serum TF approaches within 6A of the iron center and has been implicated by labelling experiments to undergo a change in reactivity upon iron binding (Shewale and Brew,1982). The equivalent residues in the N lobe of hLF and C lobe of OTF have been replaced by arginine and glutamine respectively, though the functional implications of these substitutions has yet to be demonstrated. The isolation of iron-binding fragments from the N lobe which correspond to Domain Nil has been reported for several transferrin molecules (Evans and 56 Madden, 1984; Legrand et al., 1984). As these fragments lack the Asp61 ligand, a plausible sequence for ligand binding might involve initial binding by the carbonate dianion and the other three protein ligands in a structurally open conformation after which the two domains would be brought within close proximity by pivoting about the flexible hinge formed by the interdomain strands (Baker et al., 1987). Coordination of the final position by Asp61 would complete the binding and lock the molecule into a tight configuration, in agreement with studies which show a significant compaction of the molecule upon iron binding. In order to understand better the structural requirements for iron-binding by T F , a study was undertaken to express recombinant molecules for the purpose of employing the techniques of site-directed mutagenesis. Previous investigators have expressed the chicken O T F gene in transgenic mice (McKnight et al., 1983) and production of a portion of rat T F fused to /J-galactosidase has also been reported (Aldred et al., 1984). The study reported in this dissertation was designed to achieve a high level of protein expression in a biological system that would allow ready recovery of the recombinant product. Our initial efforts have been to produce a recombinant N lobe half molecule of human serum T F , though we are now extending our analysis in attempting to express both the C lobe half molecule and a full length T F species. Successful completion of these goals will allow us to dissect separate iron binding sites, assess the requirements for T F receptor binding and investigate individual and collective properties of the two lobes. 57 II. Experimental Procedures and Materials A. General Materials and Supplies The large fragment (Klenow) of E. coli D N A polymerase I, T4 polynucleotide kinase, T4 D N A ligase, and all restriction enzymes were purchased either from Pharmacia-PL Biochemicals or Bethesda Research Laboratories. Deoxy- and dideoxyribonucleotidetriphosphates were purchased from Pharmacia-PL Biochemicals. a-[ PJdATP and 7-[ P]ATP were from Amersham and New England Nuclear. Hen egg white lysozyme, deoxyribonuclease I, ribonuclease A and goat anti-human transferrin antiserum were from Sigma. Polyethylene glycol 4000 (PEG) was from B D H Chemicals. Ferriprotoporphyrin IX dimethylester was obtained from Porphyrin Products, Logan, U T . Ru(NH3)gCl3 (Alfa) was purified as described by Pladziewicz et al. (1973). A l l other chemicals were reagent grade or better. The three-dimensional coordinates for the crystal structure (1.9A resolution) of the trypsin-solubilized form of bovine microsomal cytochrome b$ were kindly provided by Dr. F.S. Mathews (Washington University, St. Louis, MO). Coordinates for rabbit serum T F (3.3A resolution) were provided by Dr. Peter Lindley (Birkbeck College, University of London, London, U.K.) . Al l imaging was performed with the M M S (molecular modelling system) package (S. Dempsey, University of California, S.D.) in the Department of Biochemistry at the University of British Columbia. B. Cytochrome b5 Studies 1. Gene construction The four oligonucleotides used to assemble the gene cassette encoding the lipase-solubilized form of bovine liver microsomal cytochrome b$ (designated B1-B4) were synthesized on an Applied Biosystems 380A D N A synthesizer by Mr. Tom Atkinson (University of British Columbia). Post-synthesis purification was achieved by 58 electrophoresis through an 8% denaturing polyacrylamide gel (7M urea), followed by reverse phase liquid chromatography of the appropriately sized product through Sep-Pak CI8 cartridges (Waters Associates)(Atkinson & Smith, 1984). Concentrations were determined by absorbance at 260 nm. (1 A 2 6 0 = 33/jg/ml of single stranded DNA) . One microgram of each oligonucleotide was phosphorylated at the 5' end using A T P and T4 polynucleotide kinase (Chaconas and van de Sande, 1980). Second strand fill-in was performed using the Klenow fragment in the presence of all four deoxyribonucleotides and the resulting blunt-ended products were then cloned into Smal-cut M13mpl8 (Messing, 1983). The fidelity of the D N A synthesis step was confirmed by sequence analysis using the chain termination method on single stranded template D N A (Sanger et al., 1977). A BamHl/Hhal fragment corresponding to the sequence covered by oligonucleotides B l and B2 , and a Hhal/EcoRl fragment covering the B3 and B4 sequence were then excised from the replicative form of the appropriate clones and ligated into BamHl/EcoRl cut pUC 19 ( Messing, 1983). CaCl2 treated E. coli JM83 bacteria were transformed with the ligation products and plasmid D N A was then prepared from select clones by the alkaline lysis procedure for restriction digest analysis (Maniatis et al., 1982). The complete nucleotide sequence of the final construction was obtained by recloning a BamHl/EcoRl fragment back into both M13mpl8 and M13mpl9 followed by dideoxy sequence analysis of both strands of the gene cassette. 2. Host Selection The pUC19 plasmid encoding the lipase-solubilized form of bovine hepatic cytochrome 65 (designated pLp£>5) was transformed into CaCl2 treated E. coli host strains and grown overnight at 37° with shaking in 2 mL of Y T broth (Maniatis et al., 1982) supplemented with ampicillin to 50 /jg/mL. One hundred microlitres of each culture was subsequently used to inoculate parallel 20 mL cultures of the same medium supplemented additionally with isopropylthiogalactoside (IPTG) to ImM and grown for 59 18 hrs. Cell densities were estimated by measuring the optical density of diluted samples at 600 nm. Equivalent volumes of cell culture were pelleted by centrifugation, resuspended in NaDodSC^j-PAGE loading buffer, boiled for 5 minutes and then resolved on a 15% NaDodSC>4 polyacrylamide gel (Laemli, 1970). The gel was then fixed, stained with Coomasie Brilliant Blue R250 (Merck) and scanned using an L K B Ultroscan X L laser densitometer. 3. Induction Studies Parallel 50 mL Y T broth samples were inoculated with one mL of overnight culture of RR1 and J M 83 strains of E. coli and grown with shaking at 37°. RR1 cultures were induced with the addition of IPTG to 1 m M and 1 mL samples were then removed from the cultures at various time points for analysis by NaDodS04~PAGE. 4. Protein Purification Transformed J M 83 bacteria containing pLp&5 were grown for 16 hours at 37° with shaking in Y T medium (4L) containing 50 jUg/mL ampicillin. This culture was used to inoculate 50 L of the same medium in a Bilthoven fermentation apparatus. This culture was grown with vigorous aeration until the O.D.600 reached 2.5, at which time the air supply was reduced to a minimum. The culture was maintained under these conditions for a further 20 hours. This latter step improved the yield of holo-cytochrome 65, presumably by permitting heme synthesis to accommodate the large load of apocytochrome 65 (S.G. Sligar, University of Illinois; personal communication). Bacteria were harvested with a C E P A continuous flow centrifuge, and the resulting cell paste was resuspended in IL of lysis buffer containing 3 mg/ml lysozyme as described (von Bodman et al., 1986). Following a one hour incubation at 37°, cell lysis was completed by a single pass through a French press (1200 p.s.i.), and the cell debris was cleared by centrifugation at 8600 x g for 30 minutes at 4 ° . KC1 and polyethyleneglycol 60 (PEG 4000) (final concentrations 0.4 M and 6% respectively) were added to the supernatant solution. This solution was stirred for 15 minutes and then centrifuged at 8600 x g for 30 minutes at 4°. The resulting supernatant solution was brought to 50% ( N H ^ S C ^ saturation, stirred at 4° for 1 hr, and then centrifuged at 8600 x g for 30 minutes at 4°. The supernatant solution was dialyzed exhaustively against de-ionized water until the conductivity equalled that of sodium phosphate buffer (20 m M , pH 7.2). The dialysate was clarified by centrifugation at 8600 x g , loaded onto a column (2.5 x 12.5 cm) of DE-52 cellulose (Whatman), washed with one column volume of 50 m M sodium phosphate (pH 7.2), and eluted with 150 m M sodium phosphate buffer (pH 7.2). Red coloured fractions were concentrated by ultrafiltration (Amicon Y M - 5 and Centriprep-10), exchanged into 20 mM sodium phosphate (pH 7.2), and loaded onto a column (5 x 150 cm) of Sephadex G75 superfine resin (Pharmacia). The column was developed with 20 m M sodium phosphate buffer (pH 7.2), and the A412.5/A28O ratio of individual fractions was determined. Fractions with an A412.5/A280 ratio of greater than 5.0 were concentrated by ultrafiltration and exchanged into 20 m M triethanolamine (pH 7.3). Further purification was achieved by chromatography on an HR 10/10 Mono-Q F P L C column (Pharmacia) in 20 m M triethanolamine (pH 7.3) with a linear gradient of 0.2 to 0.29 M NaCl. The S64A mutant form of the bovine cytochrome was not stable when chromatographed on Mono-Q F P L C . Therefore, concentrated fractions from the first Sephadex G75 column were re-chromatographed on a second smaller column (2.5 x 100 cm) of Sephadex G75. For all forms of the cytochrome, fractions with an A412.5/A28O ratio of >5.9 were concentrated by ultracentrifugation and stored in liquid nitrogen. 5. Dimethylesterheme IX Substitution Replacement of the native protoheme IX group with dimethylesterheme IX (DME) was performed as described previously (Reid et al., 1984). Briefly, purified 61 cytochrome 65 was exchanged into de-ionized water, concentrated to 10 mg/mL by ultracentrifugation, chilled on ice and acidified to pH 1.5 with HC1. Al l subsequent manipulations were performed at 4°. This solution was extracted with an equal volume of methylethylketone four times and then dialyzed consecutively against: 1. 50 mg/L NaHCG-3, 320 mg/L E D T A 2. 50 mg/L NaHCC-3 3. 20 m M sodium phosphate pH 7.2 4. 50 mg/L NaHCOs. D M E was dissolved in dimethylformamide and added to the dialyzed apocytochrome dropwise. The reconstituted DME-cytochrome was separated from the free D M E by chromatography on a column (100 X 2.5 cm) of Sephadex G75 (Pharmacia). 6. Tryptic Peptide Mapping Apocytochrome 65 was prepared by the method of Teale (1959; Reid et al., 1984). One milligram of the apoprotein was digested in 0.5 mL of 0.2 M N H 4 H C O 3 with 12.5 fil of a stock solution of L-l-tosylamide-2-phenylethyl chloromethyl ketone (TPCK)-trypsin (Worthington, Cooper Diagnostics)(1.0 mg/mL TPCK-trypsin in 1 m M HC1) at 37° for 6 hrs. An additional 12.5 /il of the TPCK-trypsin stock was then added, and the hydrolysis continued for a further 18 hours. After the addition of 100 /zl of 1M HC1, the sample was lyophilized, and the hydrolysis procedure was repeated. The final lyophilized pellet was dissolved in 1.0 mL of 0.05% trifluoroacetic acid (TFA). The remainder of the tryptic mapping procedure was performed by Marcia Mauk (Dept. of Biochemistry, University of British Columbia). Reverse phase H P L C tryptic peptide maps were developed from 100 /zl samples of this digest with a Varian Model 5060 H P L C system equipped with a Vista 401 data station and fitted with an Alltech CI8 reverse phase column (4.6 x 250 mm). The gradient (all solutions in 0.05% T F A ) was linear, 0-60% acetonitrile, over 135 minutes at a flow rate of 1.0 mL/min. 62 7. Amino Acid Sequence Determination The amino-terminal sequence of recombinant lipase solubilized cytochrome 65 and the total sequences of selected tryptic peptides obtained from H P L C maps of the authentic trypsin-solubilized bovine liver cytochrome 65 were determined on 1 nmol samples of each. Vapor phase amino terminal sequence analysis of tryptic peptides was performed with an Applied Biosystems 477A Protein Sequencer by the Protein Microchemistry Centre at the University of Victoria. Amino-terminal sequence determination of the recombinant cytochrome was performed by Prof. Ian Clarke-Lewis at the Biomedical Research Centre at the University of British Columbia. 8. Limited Tryptic Hydrolysis Tryptic hydrolysis to convert the recombinant lipase-solubilized form of cytochrome 65 to the trypsin-solubilized form was performed by the method of Strittmatter and Ozols (1966). Briefly, purified protein was exchanged into 20 m M sodium phosphate (pH 7.5), TPCK-trypsin was added to a final concentration of 50 -/ig/mL and the digestion allowed to proceed for two hours at 25°. Tryptic hydrolysates were chromatographed on a Mono-Q column (HR5/5)(Pharmacia) in 20 m M triethanolamine (pH 7.3) using a linear gradient of 0.19 to 0.32 M NaCl. 9. Site-Directed Mutagenesis Modifications to the gene encoding the lipase-solubilized form of bovine cytochrome 65 were accomplished with three separate techniques. The correction of codons for positions 11,13 and 57 of the recombinant Lpb$, the substitution at position 64 and the production of a gene encoding the trypsin-solubilized form of bovine cytochrome £5 was accomplished by the method of Zoller and Smith (1983) with single stranded M13mpl8 template D N A purified from RZ1032 strain E. coli, using the dut-, 6 3 ung- selection procedure (Kunkel, 1985). The alteration of the lipase-solubilized 65 gene cassette to encode the erythrocytic form of human cytochrome 65 was accomplished using the Polymerase Chain Reaction (PCR) based procedure of Nelson and Long (1989). The production of the E11Q,Q13E and N57D forms of bovine lipase-solubilized 65 was achieved by splicing BamHl/Accl and AccI/EcoRl restriction fragments from the triple mutant and corrected wild-type gene cassettes. 10. Spectroelectrochemistry Potentiometric titrations were performed with an optically transparent thin layer electrode ( O T T L E ) described previously (Reid et al., 1982,; Reid 1984) with Ru(NH 3)6Cl 3 used as a mediator. Solution potentials were measured against the saturated calomel electrode (Radiometer #K4112)(SCE) and converted to the hydrogen scale as described by Dutton (1978). Spectra were processed and Nernst plots obtained using OLIS software (On-Line-Instrument-Systems, Jefferson, GA) . Al l reported midpoint potentials represent at least two independent titrations. C. Human Serum Transferrin Studies The majority of the studies on the expression of human serum transferrin were done in collaboration and consultation with Drs. Robert C. Woodworth and Anne B. Mason (University of Vermont). 1. Isolation of Human Serum Transferrin cDNA The initial isolation of a full-length cDNA for human serum transferrin was performed by Mrs. Susan Lejay. A human liver cDNA library constructed in the E. coli expression vector pKT-218 (Prochownik et al., 1983) was kindly provided by Dr. Stuart Orkin (Harvard University). This cDNA library was screened by using a synthetic oligonucleotide coding for the amino-terminal eight amino acids of serum hTF as a 64 hybridization probe. This oligonucleotide corresponded to nucleotides 88-111 of the human serum T F cDNA sequence reported by Yang et al. (1984). The oligonucleotide XI was end-labelled with T4 polynucleotide kinase and i-[ P]ATP (Chaconas & van de Sande,1980). Replica nitrocellulose filters lifted from approximately 105 colonies (Schleicher and Schuell) were screened under high stringency conditions (final washes in IX SSC, 1% NaDodSC>4 at 60°). Restriction endonuclease mapping of positive clones and D N A sequence analysis of positive clones was performed using standard procedures with pUC19 and M13mpl9 vectors, respectively (Maniatas et al., 1982; Messing,1983; Sanger et a/.,1977). 2. Expression Vector Assembly A Pstl/Hhal restriction fragment spanning the coding region for the signal sequence and N-terminal lobe of human T F was sub-cloned into M13mpl8. Oligonucleotides were synthesized and purified as described for the cytochrome b5 methods above. Subsequent alteration of this sequence was accomplished by using a mutagenesis kit purchased from Amersham and the procedures described by Taylor et al. (1985). Unique Hindlll and Kpnl endonuclease recognition sites and two translational stop codons were engineered to allow the excision of coding sequence for the mature N -terminal lobe of T F . A n altered version of the expression vector pKK223-3 (Brosius and Holy, 1984), termed ApKK.223-3 and oligonucleotides encoding the E. coli alkaline phosphatase signal sequence were kindly provided by Dr.'Deborah Cleveland (University of British Columbia). The expression vector p N U T (Palmiter et al.,1987) was kindly provided by Dr. Richard Palmiter (Howard Hughes Medical Institute, University of Washington). A HpaW/Hin&lU restriction fragment corresponding to the signal peptide and N-terminal lobe of human T F was made blunt using the Klenow fragment and deoxyribonucleotides and then cloned into the large Sma\ fragment of p N U T . Al l subsequent manipulations of D N A were performed using standard methodologies 65 (Maniatis et al., 1982). 3. DNA Transfection and Eukaryotic Cell Culture Baby hamster kidney (BHK) cells were a gift of Dr. Richard Palmiter (Howard Hughes Medical Institute, University of Washington). Super-coiled plasmid D N A was purified from E. coli JM105 by two successive centrifugation steps through cesium chloride gradients. B H K cells were grown in Dulbecco's modified essential medium ( D M E M ) (Gibco) containing 10% fetal calf serum (Gibco) and 50 U / m l penicillin G and 37 U / m L streptomycin (both from Sigma) in a 37° humidified incubator under 5% OO2. For the transfection procedures, approximately 107 cells were seeded per 90-mm culture dish and 12 hours later were transfected with 10 ng of plasmid D N A using the calcium phosphate co-precipitation procedure as described by Searle et al. (1985). After 24 h, the medium was changed to medium consisting of a 1:1 (v:v) mixture of D M E M and D M E M harvested from semi-confluent B H K cultures (conditioned D M E M ) , with a final concentration of 10% fetal calf serum and 500 methotrexate (MTX). Stocks of selected cells were stored in liquid nitrogen using D M E M containing 20% fetal calf serum and 10% dimethylsulfoxide (DMSO) (Sigma, tissue culture grade) as a preservative. Large-scale roller cultures were initiated by seeding approximately 5 x 10 cells into each 850 cm 2 roller bottle (Corning) containing 100 mL of medium consisting of a 1:1 (v:v) mixture of D M E M and Hamm's F-12 nutrient mixture, with a final concentration of fetal calf serum of 5% and supplemented with 500 fiM M T X . Cultures were induced at 80% confluency by the addition of ZnSCu to a final concentration of 0.02 mM. The medium was harvested at 48 h intervals and replaced with 100 mL of the same medium. 66 4. Immumoprecipitation and Western Blotting Immunoprecipitation of cell culture medium and cell lysates was performed by the method of van Oost et al. (1986). Briefly, medium was removed from Zn-induced cultures, and the cell layer rinsed twice with phosphate buffered saline (PBS). The cell layer was solubilized with a detergent solution containing 0.15 M NaCl, 20 m M Tri s -HCl (pH 8.0), 0.5% sodium deoxycholate (NadChol), 1% Nonidet-P40 and 0.1 % NaDodSC^. The cell debris was removed by centrifugation at 8000 x g at ambient temperature. Two mL of a detergent solution containing 0.5% NadChol, 0.1% NaDodSO^, and 1% Nonidet-P40 was added to each 10 mL of cell culture medium harvested. Both cell culture medium and cell lysates were treated with pre-immune mouse serum (1000-fold diluted) and 1 mL of formalin-fixed Staphylococcus aureus cells (10% v:v)(Bethesda Research Laboratories) and then cleared by centrifugation at 8000 x g. Goat anti-hTF serum (Sigma) was then added (250-fold dilution), left overnight at 4° and immunoconjugates were then recovered with the addition of S. aureus cells. Precipitated material was resolved by electrophoresis on 12% polyacrylamide gels in the presence of NaDodSC>4 (Laemmli, 1970), followed by electroblotting onto nitrocellulose membranes (Schleicher & Schuell). Blots were incubated in PBS containing 0.1 mg/mL gelatin, then treated with goat anti-hTf antiserum (250-fold dilution), and finally developed with an alkaline phophatase conjugated, rabbit anti-goat IgG antibody (Promega) according to the supplier's instructions. 5. Protein Purification Harvested culture medium was made 0.01% in phenylmethylsulfonylfluoride (PMSF)(Sigma), and sufficient ferricnitrilotriacetic acid (Fe I J I(NTA)2) was added to saturate all transferrin in the medium. After stirring for 30 min, the solution was dialyzed for 24 h versus cold running tap water and then for 12 h versus several changes of cold de-ionized water. Concentrated Tris-HCl buffer, pH 8.4, was added to a final 67 concentration of 5 mM. The preparation was centrifuged at 6000 x g to remove insoluble debris, and then loaded onto a column (2.5 x 80 cm) of DEAE-Sephacel (Pharmacia) equilibrated with 10 m M Tris -HCl buffer, pH 8.4. The column was eluted with a linear gradient of NaCl (0-0.3 M) in the same buffer. Fractions showing pink colour were analyzed by NaDodSC»4-PAGE and fractions containing the recombinant protein were pooled and concentrated by ultrafiltration over an Amicon PM-10 membrane. The protein was then chromatographed on a column (2.5 x 100 cm) of Sephadex G-75 Superfine (Pharmacia) equilibrated with 100 mM ammonium bicarbonate and fractions were analyzed by NaDodS04-PAGE. Fractions still containing full-length bovine transferrin and the recombinant protein were re-chromatographed on the same G-75 column. The recombinant protein was purified to homogeneity by F P L C on a column (1 x 10 cm) of Polyanion SI (Pharmacia) using a linear gradient of NaCl (0-0.3M) in 50 m M Tr i s -HCl , pH 8.0, over a period of 1 h at a flow rate of 1 mL/min. Fractions of 1 mL were collected. NaDodSC>4-PAGE was performed on 5-12% gradient gels, and urea-PAGE was performed according to a modification (Brown-Mason & Woodworth, 1984) of the Makey & Seal procedure (1976). Iron was removed from the holoprotein by incubation in a buffer containing ImM N T A , ImM E D T A , and 0.5M sodium acetate, pH 4.9. For iron-binding titration studies, the apoprotein was made 10 mM in NaHCC>3 and titrated with F e m ( N T A ) 2 while monitoring the absorbance at 465 nm. 6. Amino-Terminal Sequence Analysis The amino-terminal sequences of both the major and minor forms of the recombinant h T F / 2 N were determined on an Applied Biosystems 470A protein sequencer in the Given Analytical Facility at the University of Vermont. 68 7. hTF/2C-pNUT Assembly The initial assembly of the hTF/2C coding sequence was accomplished by Dr. Janet Lineback (Florida International University). A construction encoding a h T F / 2 C recombinant species was produced using a PCR procedure to effect the splicing of the natural signal peptide sequence to the C lobe coding sequence. The oligonucleotide primer defining the 5' end of the clone encoded the natural signal sequence of the hTF c D N A and residues 334-341 of the C-terminal lobe and produced a Smal site at the 5' terminus of the amplified product, while the 3' oligonucleotide spanned a region of the 3' non-coding sequence and introduced an additional Smal site. The initial pKT218-hTF c D N A vector was first cut with EcoKl to remove almost the entire coding sequence for the N-terminal domain, after which the remaining coding sequence served as a template for 25 rounds of PCR amplification with Thermophillus aquaticus (Taq) D N A polymerase (AmpliTaq; Perkin-Elmer) using reaction conditions recommended by the supplier. The amplified product was cut with Smal, cloned into Smal-cut p N U T vector and introduced into B H K cells as before. In order to determine the fidelity of the PCR amplification procedure, the Smal-cut PCR amplification product was also cloned into the Smal site of the vector BlueScript KS+ (Stratagene) and subject to ExoIII deletion analysis (Heinikoff, 1984). Clones from individual time points were analyzed by restriction digest and then infected with helper phage K107 to produce single stranded template for dideoxy sequence analysis. 69 III Results A. Cytochrome D5 Studies 1. Gene Design and Synthesis The assembly strategy for the gene cassette encoding the lipase-solubilized bovine microsomal cytochrome 65 protein is shown in Figure 17. The oligonucleotide pairs (B1/B2 and B3/B4) were designed to overlap sufficiently so as to be stable at the extension temperature (37°). A n autoradiogram of the extension of end-labelled primer pairs is shown in Figure 18. Both primer pairs could be extended to yield appropriately sized products of 170 bp., with yields approaching 50% based on comparison of band intensities of the extension products and remaining unextended oligonucleotides. The sequence of the oligonucleotides was screened prior to synthesis to ensure that the regions of overlap between pairs would constitute a unique priming structure. Nevertheless, the extension reaction with oligonucleotides B3 and B4 produced a substantial amount of a shorter product of approximately 140 bp. As this reaction product was seen only when oligonucleotide B4 was labelled, it must represent an alternate priming event in which the 3' end of B4 hybridizes to an internal sequence in B3. Such illegitimate priming may restrict the number of oligonucleotides that could be used simultaneously in fill-in synthesis procedures. In the present study, no full-length product was observed in reactions where all four oligonucleotides were annealed and extended (results not shown). A second limitation of this strategy is the fidelity of the polymerase activity. The first clone sequenced during assembly contained a deletion of a single thymidylate residue, and resulted in the introduction of an ochre codon at position 29 of the translation product. The use of alternate polymerases with greater processivity, such as that of T7 bacteriophage (Tabor and Richardson, 1987), may increase the accuracy of second strand synthesis. Bl B2 B3 B4 1 1. Klenow/dNTPs Kpnl, I •—L 1 I  iHhal 1. Blunt end clone into M13mp l8 2. Sequence 3. Release with restriction digest Hhal r i i I EcoRl ' i i i 1 1. Ligate into pUC 19 2. Transform E. coli JM83 Figure 17: Assembly Strategy for the Cytochrome D5 Gene Cassette Two pairs of overlapping oligonucleotides (B1/B2 and B3/B4) were annealed, then filled-in using Pol I (Klenow). The resulting blunt fragments were ligated into M13mpl8, sequenced, then released with the indicated restriction endonuclease digestions. These fragments were then cloned directionally into pUC19 and transformed into E. coli JM83. Internal Kpnl, Hhal and EcoRl restriction sites were included in the design of the gene cassette to allow the directed assembly of fill-in reaction products into pUC19. Direct cleavage of the extension products proved to be inefficient, even though 71 EXTENSION PRODUCTS B1 B2 B3 B4 154 bp. LABELLED OLIGOS 75 bp-Figure 18: Second Strand Fil l- in Reaction of B1-B4 Oligonucleotides Four separate oligonucleotide extension reactions were performed using paired oligonucleotides, one of which was end-labelled with 3 2 P . Reaction products were electrophoresed through a denaturing polyacrylamide gel (10% polyacrylamide, 8M urea). The indicated size markers are 3 2P-labelled Hinfl digestion products of pBR322. additional bases had been allowed at the presumptive termini of the extension products to encourage endonuclease recognition. Thus the extension products were cloned without modification into the Smal site of M13mpl8 for dideoxy sequence analysis and were 72 Tp F 10 B 1 GTACGGTACJCGTA? AGGA 3 AAC7AAC AAT G AGC AAAG C T G T C AAAT AC T AC AC T C T GG AAG R B S M S KTA V K CATG(^ATGGCATTTCCTCTTGTTGTTACTCGTTTCGACAGTTTATGATGTGAGACCTTC Kpn I 20 30 E I Q K H N N S K S T W L I L H Y K V Y AAATCCAAAAACACAACAACTCGAAATCGACGTGGCTGATCCTGCACTACA7AAGTATACG TTTAGGTTTTTGTGTTGTTGAGCTTTAGCTGCACCGACTAGGACGTGATGTTTCATATGC 40 50 D L T K F L E E H P G G E E V L R E Q A ACCTGACTAAATTCCTGGAAGAGCACCCGGGAGGCGAAGAAGTCCTGCGCGAACAGGCCG B2 TGGACTGATTTAAGGACCTTCTCGTGGGCCCTCCGCTTCTTCAGGACGCGCTTGTCCGGC 60 70 G G D A T E N F E D V G H S T D A R E L B3 GCGGTGACGCGACTGAAAACTTTGAGGATGTTGGCCATAGTACCGACGCTCGTGAACTGT CGCCACTGCGCTGACTTTTGAAACTCCTACAACCGGTATCATGGCTGCGAGCACTTGACA 80 T , P 90 S K T F I I G E L H P., D D . R y S K I T K P CiSAAAACGJLTCATCATCGGTGAGCTGCACGCGGACGATCGTTCTAAAATCACCAAACCGT GCTTTTGCAAGTAGTAGCCACTCGACGTGGGCCTGCTAGCAAGATTTTAGTGGTTTGGCA S E S # # CCGAATCTTAATGAGJAATTCGTCTGA GGCTTAGAATTACTCTTAIJGCAGACT N . Eco RI 73 Figure 19: Nucleotide Sequence of the Lipase-Solubilized, Bovine Microsomal Cytochrome D5 Gene Cassette. The oligonucleotides used in the synthesis of the gene cassette (B1-B4) are underlined. Restriction endonuclease recognition sequences used in the final assembly step are indicated by bent arrows while the sites of trypsin cleavage which yield the heme-binding core domain are shown by straight arrows. Termination codons for the lac Z (*) and cytochrome 65 (##) translation products are indicated while the ribosome binding site for cytochrome 65 initiation is boxed. then efficiently released from the replicative form and assembled into pUC 19 (see Figure 19). The gene cassette utilizes a transcriptional fusion strategy with the lacZ transcript from pUC and includes an in-frame stop codon for the lacZ translation product that is followed immediately by a consensus ribosome binding (Shine-Delgarno) site (Figure 20). The spacer region between the ribosome binding site and initiator methionine was modelled on the Pseudomonas cytochrome P-450cam sequence, which has been used successfully in the expression of a synthetic gene for rat liver cytochrome b$ in E. coli (von Bodman et al., 1986). This region was modified to conform to consensus base rules for this region (Sherer et a/., 1980; Stormo et al., 1982) by shortening the distance between the ribosome binding site and initiator methionine to 9 bp. Codons were chosen to reflect the E. coli bias (Maruyama et al., 1986). 4. Expression Studies Transformation of E. coli JM83 with the pUC construction containing the gene for the lipase-solubilized cytochrome b$ (pLp^) resulted in high levels of the recombinant protein as demonstrated by the bright pink colour of bacterial pellets. As a first approach towards optimizing expression, several different strains of E. coli were transformed, and the level of recombinant protein in each was estimated by NaDodSC>4-PAGE analysis. A representative trial is shown in Figure 21. As judged by laser RBS MCS UAA ' M e t b c lac Plac ° l a c J b5CODING L r 1 UAA Transcriptional UAG Termination | Signals > Met lac Kpnl RBS t EcoRl I Induction/Transcription T UAA UAG Figure 20: Transcriptional Fusion Strategy for the Expression of Recombinant Cytochrome b5. The cytochrome 65 gene cassette (large block) includes an in-frame translational termination codon for the lac Z product (UAA) . Induction of the lac promoter of pUCl9 (Pi a c ) results in a polycistronic message, thus the information for cytochrome 65 expression is "transcriptionally fused" to the lac operon. Upon translation, the lac Z product is terminated, and the translation machinery re-initiates at the adjacent ribosome binding site (RBS65) after which the cytochrome 65 coding information is translated. scanning densitometry, the recombinant cytochrome 65 accounts for up to 15% of the total stain density (Figure 20, sample RR1). To optimize the expression of recombinant protein further, an induction time course experiment was performed. Several cultures of transformed E. coli RR1 were induced at varying times after which samples were removed for optical density 75 DH5 JM83 JM101 M.W .STANDARDS RR1 OCH (kDa) Figure 21: E . coli Host Selection for Cytochrome D5 Expression Plasmid pLpi?5, encoding the lipase-solubilized form of cytochrome b$ was used to transform four different strains of E. coli. DH5, RR1 and JM101 strains were grown in Y T medium supplemented with 1 mM IPTG, while JM83 strains were grown in Y T medium alone. Based on O.D.600nm readings, equivalent amounts of bacteria from 16 h. cultures were lysed by boiling and the resulting supernatants were analyzed by NaDodS04-PAGE (15% polyacrylamide). Also included was an overloaded amount of JM83 culture transformed with an ochre mutant of pLpZ>5. Cytochrome 65 runs anomalously large in these gel systems ( M r 14 kDa versus an empirically estimated weight of 10,590 Da for apo-Lp^) . determination and NaDodSGvPAGE analysis. As shown in Figure 22, the total amount of recombinant protein was highest when IPTG induction was initiated early in the culture (compare lanes A and J), and the maximal yield of all cultures was observed only Figure 22: IPTG Induction Time Course of Lpb$ Expression Parallel cultures of pLpb$-transformed E. coli RR1 and JM83 were grown in Y T medium. RR1 cultures were induced with IPTG (1 m M final concentration) at various culture densities. Samples were harvested at subsequent time points for analysis by N a D o d S 0 4 - P A G E (15% polyacrylamide). Lane Host O.D.gQO a t Induction O.D.ggo a t Harvest A J M 83 - 2.5 B J M 83 - 1.5 C RR1 2.0 2.5 D RR1 2.0 2.1 E RR1 1.5 2.5 F RR1 1.5 1.9 G RR1 1.0 2.5 H RR1 1.0 2.2 I RR1 1.0 1.3 J RR1 0.5 2.5 K RR1 0.5 1.6 L RR1 0.5 0.9 77 after an 18 hour incubation. This observation demonstrates that the bacteria continue to produce recombinant protein well after entering stationary phase (O.D.6oo=2.4). Fifty liter fermentation runs with JM83 routinely yield up to 500 mg of purified protein. However this yield represents only about 10% of the expected yield based on the gel analysis described above, if it is assumed that 15% of the wet weight of E. coli is protein (Lehninger,1970). As this discrepancy may reflect an inability of the bacterial host to supply sufficient heme for the cytochrome load, experiments were undertaken in which exogenous heme was added to culture lysates in an attempt to saturate any apoprotein which might remain in a relatively native form. No significant increase in protein yield was observed in these experiments, nor in other studies in which porphyrin precursors such as glycine and 5-aminolevulenic acid were added to the cultures. Alternatively, this discrepancy might suggest that only a fraction of the recombinant protein is in a form competent to bind heme, whether as a result of intracellular compartmentalization or protein degradation. Anomalous staining of cytochrome 65 in the gel analysis could also explain this finding. 5. Amino-terminal Sequence Analysis Gas phase protein sequencing of the amino-terminal region of the recombinant cytochrome showed complete agreement with the predicted sequence from D N A analysis (Table I). The absence of a peptide containing formyl-methionine in the tryptic peptide maps (see section IIIA7. Tryptic Peptide Mapping) indicates complete processing of the amino-terminus as reported for recombinant rat liver cytochrome 65 (von Bodman et a/.,1986). 6. FPLC Analysis of Authentic and Recombinant Cytochromes b$ Our efforts to characterize fully the recombinant cytochrome 65 included performing limited tryptic digests on the holoprotein in order to convert the lipase-78 Recombinant Cytochrome b Tryptic Peptide T2 Tryptic Peptide T7 uncorrected authentic authentic uncorrected Cycle residue yield* residue yield* residue yield* residue yield* 1 Ser 309 Tyr 818 Glu 963 Glu 989 2 Lys 156 Tyr 897 Gin 751 Gin 724 3 Ala 240 Thr 306 Ala 786 Ala 901 4 Val 809 Leu 572 Gly 612 Gly 655 5 Lys 180 Glu 545 Gly 725 Gly 875 6 Tyr 664 Glu 585 Asp 504 Asp 568 7 Tyr 544 lie 299 Ala 510 Ala 599 S Thr 380 Gin 434 Thr 242 Thr 252 9 Leu 145 Lys 188 Glu 375 Glu 473 10 GIu 135 Asp 312 Asn 396 11 . Gin 583 Phe 324 Phe 414 12 He 359 Glu 275 Glu 370 13 Asp 238 Asp 277 14 Val 183 Val 239 15 Gly 152 Gly 239 16 His 78 His 106 17 Ser 54 Ser 54 18 r Thr 54 Thr 55 19 Asp 83 Asp 102 20 Ala 61 Ala 78 21 Arg 10 Arg 15 *pmol 79 Table I: Amino-Terminal Sequence Analysis of Recombinant Lpbg. Gas phase sequence analysis was performed on the initial recombinant Lpb^ (Ql 1,E13,D57), on tryptic digestion peptide T2 from the authentic tryptic core protein from bovine liver microsomes (corresponding to residues 6-14, using the Lpb$ recombinant numbering system) and on peptide T7 (residues 52-76) from both the recombinant Lp£>5 and authentic tryptic core proteins. solubilized form to the corresponding trypsin-solubilized form as reported for the authentic bovine protein (Ozols and Strittmatter,1966). This process releases peptides from both the N - and C-termini (see Figure 19) and therefore allowed direct comparison of recombinant and authentic molecules of identical chain length. F P L C analysis of the resulting digestion products revealed three species (data not shown), none of which exhibited elution behavior identical to that of the authentic trypsin-solubilized cytochrome b$. This profile persisted even when the heme group was removed and subsequently replaced with exogenous hemin, indicating that the altered tryptic susceptibility was not the result of defective protein folding. The major digestion product had a prolonged retention time relative to that of the authentic sample (compare panels A and D, Figure 23), suggesting that it is more anionic in nature. 7. Tryptic Peptide Mapping The initial peptide map of the recombinant cytochrome indicated that all of the predicted tryptic fragments were present. Within the resolution and error of the mapping process, the H P L C retention times of these peptides appeared to match those of the authentic trypsin-solubilized cytochrome. Additional peptides in the profile of the recombinant sample were predicted to result from the amino- and carboxy-terminal extensions of the lipase-solubilized recombinant form versus the trypsin-solubilized bovine liver cytochrome b$. Amino acid composition analysis of these additional peptides confirmed this expectation. 08 81 Figure 23: FPLC Elution Profiles of Recombinant and Authentic Microsomal Cytochrome D5 Limited tryptic digestion was performed on the corrected (El 1,Q13,N57) and initial forms (Ql 1,E13,D57) of recombinant Lpb$ (see IIB8, Experimental Procedures and Materials). Panel A: Corrected recombinant Lpb$ ( V e 15 mL) and tryptic core of authentic bovine microsomal cytochrome 65 ( V e 25 mL). Panel B: Tryptic core of corrected recombinant Lp&5 (280 nm). Panel C: Tryptic core of corrected recombinant Lpb$ and the authentic bovine microsomal cytochrome 65 (405 nm). Panel D: Major product from the limited tryptic digestion of initial recombinant Lp&5 (Q11,E13,D57). Comparisons of the electrochemical (vide infra), chromatographic, cytochrome c-binding and methemoglobin-binding behavior (M.R. Mauk, W.D. Funk, unpublished results) and N M R properties (Burch et al., 1988) of the recombinant versus authentic cytochromes showed additional functional differences between the two proteins. Circular dichroism (CD) spectroscopy of the recombinant protein showed no significant difference relative to the authentic trypsin-solubilized protein (data not shown), suggesting that no gross structural re-organization of the molecule had occurred. However multiple peptide mapping experiments of both the recombinant and authentic cytochromes demonstrated slight but reproducible differences in the retention times of peptides T2 and T7 from the two proteins (Table II). Gas phase sequence analysis of these peptides indicated that the peptides from authentic trypsin-solubilized cytochrome 65 were in concordance with the predicted amino acid sequence from c D N A analysis (Christiano and Steggles, 1989), while the peptides from the recombinant protein agreed with the sequence obtained by Ozols and Strittmatter (1969), upon which the gene cassette sequence had been targetted (Table I). The two sequences differ in the amidation status of three residues; the corrected assignments are: Asn57, G l u l l , and Gin 13 (numbering of Ozols and Strittmatter, 1969) 82 Retention Time (min) Peptide Residues Authentic3 Recombinant'3 Difference 5-6 — 6.2 ~ ~ -T1 7-9 18.2 18.4 0.2 T2 10-18 71.4 72.8 1.4 T3 19-23 16.5 16.4 0.1 T4 24-32 109.0 109.1 0.1 T5 33-38 55.0 54.8 0.2 T6 39-51 86.4 86.2 0.2 T7 52-76 63.9 66.0 2.1 T8. 73-76 28.4 28.5 0.1 T9 77-88 97.7 97.7 0.0 89-90 . . . 6.2 . . . 91-97 36.6 - - . aAverage of 5 peptide maps. ^Average of 11 peptide maps. Table II: H P L C Retention Times of Tryptic Peptides from Recombinant Bovine Microsomal Lipase-solubilized Cytochrome 05 (Q11,E13 ,D57) and Authentic Bovine Microsomal Trypsin-solubilized Cytochrome D5 Peptides generated from the complete tryptic digestion of recombinant and authentic cytochrome b$ samples were analyzed by H P L C . The mean retention times for each fragment are presented, along with the differences in retention times between molecular species. The authentic trypsin-solubilized cytochrome 65 is shorter at both the amino-and carboxy-terminus compared to the recombinant lipase-solubilized form. 83 8. Primary Sequence Correction The amidation status of residues 11, 13 and 57 of the bovine liver cytochrome 65 was confirmed independently by sequence analysis of the corresponding c D N A by Christiano and Steggles (1989). Two oligonucleotides spanning the sequence for residues 11 and 13 and for residue 57 were synthesized and used to effect the necessary replacements by site-directed mutagenesis. The resulting wild-type (corrected) lipase-solubilized cytochrome still displayed a different F P L C retention time than that exhibited by the authentic trypsin-solubilized protein (Figure 23, panel A). Significantly, the trypsin digestion product of the corrected protein eluted as a single peak (panel B) that eluted with a retention time identical to that of the authentic trypsin-solubilized protein (panel C). The complex elution profile of the triple mutant trypsin hydrolysis reaction suggested that the termini of the molecule were less susceptible to proteolytic hydrolysis by trypsin than wild-type lipase 65. To investigate this possibility, restriction fragments from the gene cassettes for the corrected wild-type and triple mutant cytochromes were joined to produce molecules mutated either at position 57 (D57) or 11 and 13 (Q11,E13). The resulting molecules were subjected to limited trypsin digestion and chromatographed as before. Surprisingly, both the D57 and Q11,E13 mutant proteins yielded a single reaction product (Figure 24, panel B) suggesting the behavior of the triple mutant towards limited tryptic digestion must result from a subtle interplay between the two ends of the molecule. The F P L C elution profiles of the lipase form of the D57 mutant shows that this mutation alone is sufficient to account for the increased anionic character of the triple mutant (compare Figure 24, panel C versus Figure 23, panel D). Interestingly, the lipase form of the Q11,E13 mutant appears to have a slightly altered F P L C retention time from the corrected wild-type molecule (compare Figure 24, panel o 00 u X u o 0.08 -0.04 -0 A ±s B J L l _ c J I I 1 1—1—1 D J L J u i — L 0 20 40 0 20 40 0 20 40 0 20 40 03 Elution Volume (mL) 85 Figure 24: FPLC Elution Profile of the Limited Digestion Products of Recombinant L p b 5 Species (Q11,E13 and D57) Limited tryptic digestion was performed on two species of recombinant Lpb$, Q11,E13 and D57 as described in Experimental Procedures and Materials, B8. Panel A: Corrected recombinant Lpb$ ( V e 15 mL) and the tryptic core of authentic bovine microsomal cytochrome 65 ( V e 25 mL). Panel B Recombinant Lpb$ species prior to digestion: Q11,E13 ( V e 22 mL) and D57 (V ( 28 mL). Panel C: Recombinant hpb^ D57 following limited tryptic digestion. Panel D: Recombinant Lpb$ Q11,E 13 following limited tryptic digestion. D versus Figure 23, panel B). Following conversion to the smaller tryptic core, the Q11,E13 mutant elutes at a later point in the gradient ( V e 33 m L , Figure 24, panel D) than does the authentic tryptic core ( V e 25 mL, Figure 24, panel B). 9. Constructions for the Expression of Trypsin-Solubilized Bovine Hepatic and Human and Erythrocytic Cytochrome ftj Conversion of the original pLp&5 plasmid to encode the smaller trypsin-solubilized form of the same protein was achieved by using site-directed mutagenesis to first delete the coding sequence of Serl and Lys2 at the 5' end of the coding sequence and then to introduce a new translational stop codon immediately following the codon for Arg84. The resulting protein was recovered with similar yields as the lipase-solubilized form, exhibited identical F P L C elution characteristics with the authentic bovine-liver isolated protein (data not shown) and has a midpoint potential of -3 mV versus SHE (P.D. Barker, personal communication). The conversion of pLpb$ to encode the human erythrocytic form of cytochrome Z>5 was accomplished using the PCR-directed mutagenesis procedure of Nelson and Long (1988). As illustrated in Figure 25, the process involves two separate PCR amplification procedures, primed by two pairs of oligonucleotides. Step 1 amplifies the sequence 86 P J J - W J J  87 B C EcoRI Hindlll . MCS w w w w I I PCR Mutagenesis Oligonucleotides B 5'-P-GGA GTA CTA GTA ACC CTG GCC CCA GTC ACG ACG TTG TAA A-OH-3' (40MER) (-STRAND) C 5'-P-CAG GAA ACA GCT ATG ACC AT-OH-3' (20MER) (+STRAND) D 5'-P-GGA GTA CTA GTA ACC CTG GC-OH-3' (20MER) (-STRAND) Figure 25: PCR Mutagenesis Strategy Panel A: The method of Nelson and Long (1989) was used to generate the human erythrocytic form of cytochrome b$ from the pLp&5 gene cassette. Step 1 involves the amplification of a segment defined by the mutagenizing oligonucleotide A and a flanking oligonucleotide B. At its 3' end, Oligo B matches the multiple cloning site (MCS) flanking sequence of p U C vectors but its 5' end consists of a unique sequence. In Step 2, the product of Step 1 is used to prime a single cycle of polymerization which extends beyond the opposite flanking region. As the 3' end of the sense strand of the Step 1 product is not complementary to the template, only the mutant strand will prime polymerization. Oligonucleotides C and D are added to the reaction in Step 3 and amplification of the intervening sequence is completed. The product of Step 3 can be digested with appropriate restriction enzymes for subsequent re-cloning. Panel B: The sequences of the flanking oligonucleotides (B,C, and D) and their relative orientations at the p U C 19 multiple cloning site (MCS) are shown. As these oligonucleotides lie outside of the MCS, they can be used with either orientation of p U C vectors as well as with the M13mp series of single strand-producing vectors. 88 enclosed by the mutagenizing oligonucleotide and a second oligonucleotide hybridized to the cloning site flanking region. The double-stranded product of Step 1 is then used to prime a second round of PCR amplification spanning the remainder of the coding sequence (see Figure 25 for details). This PCR-based mutagenesis procedure offers several advantages over standard M13-based procedures. First, the amplification strategy assures that only the mutant will be produced in significant amounts, thereby eliminating the sometimes tedious process of screening multiple samples as is normally the case with M13-based mutagenesis. In addition, PCR mutagenesis is performed on double stranded template and thus eliminates both the need for isolating single-stranded template and the inherent shuttling of modified coding sequences back into the expression plasmid. The coding sequence of the resulting mutants was readily verified by double-stranded dideoxy sequence analysis (Gatermann et al., 1988) using both forward and reverse orientation universal sequencing primers. One recurrent problem with this PCR procedure has been the inadvertent introduction of additional nucleotide substitutions during the amplification procedure. These errors were found predominantly at the nucleotide pair immediately 5' to the non-priming (5') end of the mutagenizing oligonucleotide, and almost certainly result from the propensity of Taq polymerase to add an additional residue to the 3' ends of amplified products (Gelfman and White, 1990). The clones resulting from the PCR process are predominantly of the desired sequence. It may be possible to eliminate such mutants introduced by polymerase errors through the addition of a blunting procedure following Step 1 using an exonuclease activity such as that provided by the Klenow fragment. The resulting plasmid encoding human erythrocytic cytochrome b$ (phEr^) failed to produce any recombinant protein as detected by NaDodS04-PAGE in the bacterial host strain E. coli JM83. The cause of this poor expression may be different from that responsible for the poor yield observed with several of the site-specific 89 mutations performed on pLpb$; these mutants fail to incorporate heme yet show equivalent amounts of recombinant protein when compared to the wild-type standard on NaDodS04-PAGE (data not shown). As the promoter regions and transcription patterns of these two constructions should be identical, these results suggest that the human erythrocytic construction is translated with extremely low efficiency. 10. Spectroelectrochemical Studies Potentiometric titrations were performed on both the corrected and triple mutant wild-type cytochromes and on the S64A mutant in both the corrected and triple mutant backgrounds. As seen in the representative spectra of Figure 26, the recombinant cytochromes displayed electrochemically reversible behavior as had been observed for the cytochrome purified from liver (Reid et al., 1982). Such Optically Transparent Thin Layer Electrode (OTTLE) potentiometric measurements rely on the presence of a small molecule mediator to transfer electrons from a thin surface working electrode to the protein (see Appendix I for a discussion of the electrochemical methods used in this investigation). This oxidation-reduction equilibrium is monitored by measuring the electronic absorption of the solution at a wavelength where the reduced and oxidized forms of the protein exhibit different extinction coefficients. The resulting data are analyzed in terms of a Nernst plot by fitting to the Nernst equation: E a p p = E1/2 + R T / n F log ([Red]/[Ox]) In general, all of the cytochrome species examined achieved electrochemical equilibrium at a given potential within 30 minutes. The electronic absorption spectra of the S64A mutant appeared to match those of the wild-type protein closely. Isosbestic points for the wild-type protein were observed at 566, 545, 535, 514, 438, 416 and 353 nm and at 567, 545, 535, 514, 438, 416 and 351 nm for the S64A mutant, indicating that this substitution did not significantly affect the heme chromophore. The corresponding Nernst plots shown in the inset of Figure 26 were calculated from absorption readings 90 0. 8 p1 ' ' 1 1 ' 1 r 350 450 550 650 Wavelength (nm) 9 1 Figure 26: Thin Layer Spectra of Recombinant Wild-type and S64A Forms of Lpb5 Thin layer spectra were obtained using an O T T L E cell apparatus mounted in a Cary 219 spectrophotometer. The applied voltage ranged from -320 to +50 mV versus SCE. The absorbance values at 423.5 nm were used to generate Nernst plots which are presented in the inserts. Cytochrome 6 5 (120 /*M), R u ( N H 3 ) 6 2 + 3 + (12/*M), 25°C, pH 7.0 (phosphate), H= 0.1 M . A: Wild-type recombinant bovine microsomal Lpbj . B: S64A variant of recombinant bovine microsomal Lpb$. taken at 423.5 nm. The resulting midpoint potentials are listed in Table III. The midpoint potential of the corrected wild-type molecule (3 mV) is virtually identical to the value reported for the trypsin-solubilized protein isolated from liver (Reid et <a/.,1982). The S64A mutant exhibits a slightly lower potential (-4 mV). This result is somewhat surprising result in that the removal of the hydroxyl group at this position was predicted to destabilize the propionate carboxyl charge, stabilize the reduced form of the cytochrome, and thereby increase the reduction potential of the protein (see Figure 7). The midpoint potentials of both the corrected wild-type and S64A species were shifted upwards by 67 mV when reconstituted with D M E heme as had been observed for the authentic liver protein (Reid et al., 1984). The S64A substitution in the triple mutant background could only be recovered from the bacterial lysate with the addition of exogenous hemin and quickly lost heme when chromatographed over MonoQ F P L C columns. Attempts to reconstitute this species with DME-heme were unsuccessful. Cyclic voltammetry was used as an alternate method for obtaining the midpoint potentials of the D57 and Q11,E13 mutant species. This procedure cycles the protein through a range of applied potentials and relies on measurement of the current produced during reduction and oxidation to determine the midpoint potential. At low potential, the protein is essentially fully reduced. As the potential is increased, the resulting 92 Protein Prosthetic Group El/2 Nemst Slope a (mVvi.SHE) (mV) wild-type (E11,Q13,N57) (correct sequence) S64A(E11,Q13,N57) wild-type (Q11£13J357) (incorrect sequence) S64A(Q11JE13J357) wild-type (Ql 1,E13,N57) wild-type (El 1,Q130357). protoheme IX DME-protoheme LX protoheme LX DME-protoheme LX protoheme LX protoheme LX protoheme LX protoheme LX 70 63 -14 -19 57.9 61.2 59.0 56.9 59.0 62.0 -15 *25 C, 0.1M sodium phosphate, pH 7.0 Determined by using cyclic voltammetry Table III: Midpoint Reduction Potentials of Recombinant Cytochrome b5 Species. Midpoint potential were determined for several forms of recombinant Lpb$ and selected dimethylester heme (DME) replacement species. Reduction potentials were estimated either from Nernst plots of thin-layer spectra measurements or by using cyclic voltametric techniques (see also Appendix I). current first increases with rising potential, reaches a maximum, called the cathodic peak and then subsides. An anodic peak is obtained upon reversing the scan and for reversible (Nernstian) reactions, the midpoint potential lies midway between the cathodic and anodic peaks, which are themselves separated by 57/n mV ( n is the number of 93 electrons in the half reaction). However, for many real systems, such as those involving proteins, the rate of electron transfer is slow compared to the rate of diffusion and the reaction is referred to as being quasi-reversible. In such cases the peak currents are separated by a larger potential (see Appendix I for additional detail on cyclic voltammetric methods). The D57 mutant displays the same midpoint potential as the triple mutant (Table III), indicating that this substitution is solely responsible for the decreased potential of the triple mutant. In concordance with this result, the Q11,E13 mutant yielded the same midpoint potential as the corrected wild-type protein (+3 mV). B. Transferrin Studies 1. cDNA Isolation and Prokaryotic Expression Constructions Approximately 100,000 colonies of a human liver cDNA library (Prochownik et al., 1983) were screened by using a 24-base oligonucleotide corresponding to the 5' sequence of the human T F cDNA as a hybridization probe. A single positive clone was isolated whose plasmid (pKT-218/hTF) displayed the identical restriction endonuclease mapping pattern as the full-length cDNA isolated by Yang et al. from the same library (1984). Dideoxy sequence analysis of this clone showed that the 5' and 3' termini were identical to those of the full-length cDNA from Yang et al. and all subsequent sequence analysis performed during this study were also in complete agreement. A Hhal/Pstl fragment from pKT218-hTF was subcloned into M13mpl8 to produce single-stranded template for use in site-specific mutagenesis. Two translational stop codons and a unique HindlW recognition were introduced to the linker peptide coding sequence and a second oligonucleotide was used to produce a Kpn\ recognition site at the signal peptide cleavage site coding sequence (Figure 27). The predicted translation product from this sequence terminates following Asp337. These 94 Hhal I Pstl Hpall BarnHI Pstl I—dGdC -I—L 1 Isolate Hhal/Pstl Clone into M13mp18 Hhal BarnHI Pstl Oliginucleotide 1 Oligonucleotide 2 1 Hhal Kpnl BarnHI Hindlll Pstl 1 I L » I I Kpnl/Hindlll X X BamHI/Hindlll Prokaryotic Expression Eukaryotic Expression 95 B 5 - 1 1 5 Wild-type . . .G l y Leu Cys Leu Ala Val Pro Asp Lys Thr . . . ...GGG CTG TGT CTG GCT GTC CCT GAT AAA ACT . . . Oligo 1 ...GGG CTG TGT CTG GGT ACC CCT GAT AAA ACT . . . Kpnl Mutant . . .G l y Leu Cys Leu Gly Thr Pro Asp Lys Thr 334 337 340 345 Wild-type . . . A l a Pro Thr Asp Glu Cys Lys Pro Val Lys Trp Cys ...GCC CCA ACA GAT GAA TGC AAG CCT GTG AAG TGG TGT Oligo 2 ...GCC CCA ACA GAT TAA TGA AAG CTT GTG AAG TGG TGT Hindi 11 Mutant . . . A l a Pro Thr Asp * * Figure 27: In vitro Mutagenesis Strategy for the Assembly of hTF / 2 N Coding Sequence Panel A: A portion of the human serum transferrin cDNA containing the entire N lobe coding sequence was subcloned from pKT-218/hTF into M13mpl8. A single round of mutagenesis was performed using two separate oligonucleotides (1 and 2) after which clones were selected for the presence of two new restriction endonuclease recognition sites, Kpnl and Hindlll. A Kpnl/Hindlll fragment was released for use in prokaryotic expression studies, while a BamUl/Hindlll fragment was used in the assembly of the pNUT-based expression system. Panel B: The nucleotide and amino acid substitutions effected by oligonucleotides 1 and ' 2 from Panel A are presented. Oligonucleotide 1 creates a unique Kpnl recognition sequence at the signal cleavage site of hTF. This substitution results in the replacement of residues -1 and +1 in the amino acid sequence and allows the separation of the mature N lobe sequence from the signal peptide sequence. Oligonucleotide 2 introduces a unique Hindlll recognition sequence and two translational termination codons immediately 3' to the codon for Asp337. manipulations allowed the isolation of a Kpnl/Hindlll restriction fragment which lacks the signal peptide sequence and encodes the amino-terminal lobe of transferrin. This fragment was then spliced to an oligonucleotide duplex which encodes the E. coli Adaptor 97 Figure 28: Vectors for the Expression of Recombinant hTF/2N in Bacteria Two separate expression vectors were assembled for the expression of h T F / 2 N in bacterial systems. Both utilized a synthetic adaptor encoding the E. coli alkaline phosphatase (APase) signal sequence fused to the coding sequence for the mature form of h T F / 2 N . The expression vector ApKK223-3 is a modified version of pKK223-3 in which the sequence separating the ribosome binding site and initiator methionine has been shortened to 11 b.p. (D.R. Cleveland, University of British Columbia). pKK223-3 utilizes the hybrid tac promoter element (Ptac) t 0 drive transcription of sequences inserted into the multiple cloning site (MCS). Transcriptional termination is assured by the presence of the E. coli ribosomal R N A operon which includes a GC-r i ch inverted repeat termination signal (rrnTlT2). A second expression system utilizing the same transcriptional fusion strategy described for the cytochrome b5 studies (Figure 20) was assembled into the parental vector pUC18. A short adaptor oligonucleotide encoding an in-frame translational stop codon for the lac Z gene product and an adjacent ribosome binding site (RBS) was ligated with the appropriate restriction fragments encoding hTF/2N from A p K K 2 2 3 - 3 / h T F / 2 N to yield pUC18-hTF/2N. alkaline phosphatase signal peptide and was subsequently introduced into the expression vector ApKK223-3 (Figure 28). ApKK223-3 (D.R. Cleveland, University of British Columbia) differs from the parental vector pKK223-3 (Brosius and Holy,1984) in that the spacer sequence between the ribosome binding site and initiator methionine codon has been shortened to 11 bases. The same alkaline phosphatase signal peptide had been previously shown to target recombinant bovine pancreatic trypsin inhibitor to the periplasmic space (Marks et al., 1986) and we had hoped that passage of the transferrin half molecule through the bacterial membrane would encourage proper folding of the protein. However NaDodSC>4-PAGE and Western analysis of lysates from transformed bacteria failed to detect recombinant protein (results not shown). As a second attempt at eukaryotic expression, a transcriptional fusion strategy similar to that described in the cytochrome b$ studies was employed for the transferrin half molecule. As illustrated in Figure 28, an adaptor oligonucleotide duplex was assembled which encoded an in-frame translational stop codon for the lac Z product of pUC18 and an adjacent ribosome binding site. This was then joined with the alkaline phosphatase-TF/2N coding sequence and introduced into p U C 18. Again, recombinant protein was undetectable in bacterial lysates. 98 2. Eukaryotic Expression Construction The expression vector p N U T (Palmiter et al.1987) contains a mouse metallothionein I/human growth hormone gene fusion that has been shown to direct the expression of high levels of human growth hormone in transgenic mice (Palmiter et al., 1983). Important components of this expression vector include the heavy metal-inducible metallothionein promoter, pUC18 sequences to allow replication in E. coli, and a dihydrofolate reductase (DHFR) cDNA driven by the SV40 early promoter and terminated by sequences from the 3' end of the hepatitis B virus (Figure 29). The D H F R c D N A (Simonsen and Levinson, 1983) encodes a mutant form of the protein which has a 270-fold lower affinity for the competitive inhibitor methotrexate (MTX) . This allows for the immediate selection of transfected cells with high concentrations of M T X (0.5 mM), thereby abrogating the need for a recipient cell line that is deficient in D H F R . The assembly of the expression vector p N U T - h T F / 2 N is illustrated in Figure 29. A BamWl/Hin&Wl restriction fragment was first isolated from the bacterial vector A p K K 2 2 3 - 3 - h T F / 2 N and then ligated with a BamHl/Hpall fragment from the 5' end of the full length c D N A vector pKT-218/hTF into M13mpl8. This allowed the isolation of an Xbal/Hin&Wl fragment which was subsequently made blunt and inserted into the large Smal fragment of pNUT. The Xbal/Hind\\\ fragment is free of the d G / d C tail found in the c D N A vector and encodes the natural signal sequence of hTF. Transcriptional termination and polyadenylation signals are retained from the 3' end of the h G H gene in the final p N U T - h T F / 2 N construction. Transfection of B H K cells with p N U T - h T F / 2 N allows for rapid selection of cells with high concentrations of M T X . Ribonucleic acid was isolated from p N U T and p N U T - h T F / 2 N transfected cells, electrophoresed through an agarose gel in the presence of formaldehyde (Maniatis et al., 1982) and then transferred to nitrocellulose. Using an 99 dG'dC-Hhal I Met _L_L_ BamHI SIGNAL N-TERM DOMAIN PKT218-p s t i hTF cDNA C-TERM DOMAIN dGdC Isolate BamHI/Hpall From Figure 27 Hpall BamHI BamHI Met Hind II 1 BamHI 1. Ligate into Acc l/Hindl l l-cut mp18 2. Xbal/Hindlll 3. Polish ends 1. Smal Digest 2.Isolate large fragment TF/2N hGII 3' pUC 18 pUC 18 100 Figure 29: p N U T - h T F / 2 N Expression Vector Assembly The coding sequence for hTF/2N was assembled from a Hpall/BamHl restriction fragment from the 5' end of the hTF cDNA and a BamHI/Hindlll fragment from the initial mutagenesis procedures described in Figure 27. This coding sequence was released by restriction digestion and the resulting fragment was made blunt and inserted into the large Smal fragment of pNUT. Plasmid p N U T includes the promoter element of the mouse metallothionein-1 gene (mMT-1) to allow metal inducible transcription of downstream sequences and the Smal digestion fragment retains the 3' transcriptional termination and polyadenylation signals from the human growth hormone gene (hGH). A c D N A encoding a mutant form of the dihydrofolate reductase gene (DHFR) is driven by the SV40 early promoter (SV40) to provide a selectable element for eukaryotic cell expression, while bacterial replication is maintained by the pUC18 sequences. oligonucleotide to the 3' end of the h G H gene as a hybridization probe, an inducible m R N A of approximately 1.4 kbp was detected in the p N U T - h T F / 2 N transfected cells, in good agreement with the predicted size of the h T F / 2 N - h G H mRNA (data not shown). Western blot analysis was performed on both the medium and solubilized cell pellet from three cell lines. Samples were treated successively with goat anti-hTF antiserum and S. aureus cells and the resulting precipitate was then electrophoresed through a 12% polyacrylamide gel and transferred to nitrocellulose. The blot was incubated with the same goat anti-hTF antiserum and finally with alkaline phosphatase-conjugated anti-goat immunoglobulin, then developed with the appropriate substrates. As seen in Figure 30, medium and cell lysates from B H K cells and pNUT-transfected cell lines showed only the expected goat immunoglobulin bands ( M r 25 kDa and 50 kDa) and a small amount of cross reactive material common to all samples, while the p N U T - h T F / 2 N material showed the presence of a strong band at Mr 37 kDa, in excellent agreement with the calculated molecular weight of recombinant h T F / 2 N based on amino acid sequence (37833 Da). 3. Isolation and Characterization of Recombinant hTF/2N Large-scale culture of the hTF/2N cell line was performed using roller culture bottles. Radioimmunoassays of the culture medium indieated an initial concentration of 101 Figure 30: Western Blot Analysis of Immunoprecipitates from Transfected Eukaryotic Cell Cultures Samples of cell lysates (a) or medium (b) from zinc-induced cell cultures were immunoprecipitated with anti-hTF antiserum. Samples of the precipitated material were resolved by using NaDodSC>4-PAGE, transferred to nitrocellulose membrane, and developed consecutively with anti-hTF antiserum and an alkaline phosphatase-conjugated anti-IgG detection system. The p N U T and h T F / 2 N - p N U T cell lines were selected in 500 fiM M T X and all cell culture utilized D M E M medium supplemented with 10% fetal calf serum. Lane 1: B H K cells Lane 2: pNUT-transfected B H K cells Lane 3: pNUT-hTF/2N-transfected B H K cells The positions of molecular size markers (xlO Da) are indicated to the right as well as the position of the major immunoprecipitated protein of M r 37000 (<37). 10-15 /xg/mL, however continued culturing of the cells in roller bottles has resulted in levels exceeding 80 /ig/mL (R.C. Woodworth and Anne B. Mason, University of 102 22 -1 4 ^ Figure 31: Isolation and Analysis of Recombinant hTF/2N Panel A: The F P L C (Polyanion SI) elution profiles of recombinant hTF/2N and proteolytically-derived hTF/2N are shown in the upper and lower tracings respectively. Four major peaks are seen for both preparations, indicative of apo and holo forms of two distinct molecular species. Panel B: The four major components (a-d) from the polyanion SI elution of recombinant hTF/2n seen in Panel A have been analyzed by NaDodSC^-PAGE (5-12% polyacrylamide gradient). The size of molecular size standards is indicated (xlO kDa). Panel C: The major components of recombinant hTF/2N (a-d) and proteolytically-derived h T F / 2 N (e-h) were resolved by urea-PAGE. The relative position of the apo and holo forms of h T F / 2 N are indicated. 103 Protein Amino Acid Sequence Ref human serum transferrin V-P-D-K-T-V-R-W-C-A-V-S MacGillivray et al.(1983) recombinant hTF/2N (major) V-P-D-K-T-V-R-W-X-A-V-S Funk et a l . (1990a) recombinant hTF/2N (minor) V-P-D-K-T-V- Funk et a l . (1990a) 1 Twelve cycles were analyzed; no residue was identified at cycle 9, however cysteine residues were not modified 1 Six cycles were analyzed Table IV: Amino-Terminal Sequence of Human Transferrin and of the Recombinant hTF/2N Half-Molecule Approximately 200 pmol of the major and minor forms of recombinant h T F / 2 N were sequenced as described in section IIC6 of Experimental Procedures and Materials. Vermont, personal communication). The reason for this increase is not clear, though it may conceivably result from an increased cell density in long term cultures as B H K cells are only partially inhibited by cell-cell contact. Recombinant h T F / 2 N was purified by a three-step procedure with 80% recovery achieved routinely. The final Polyanion SI elution showed four distinct peaks (a-d, Figure 31, panel A) , representing the apo- and holo- forms of both a major (>95%) and minor (<5%) species. NaDodS04-PAGE analysis of the peaks a-d showed the major and minor forms to be monodisperse and of equal relative molecular size (Figure 31, panel B) and amino terminal sequence analysis showed identical sequences for both forms (Table IV). The visible absorption spectrum of iron-saturated recombinant h T F / 2 N and human serum transferrin is shown in Figure 32. Absorption maxima for both human serum T F and recombinant hTF/2N were observed at 465 nm. Spectral ratios (A28O/465) 1.0 recombinant HTF/2N Absorbance 0.6 J 0.2 400 480 X (nm) human serum TF 105 Figure 32: Visible Absorption Spectra of Recombinant hTF/2N and Human Serum Transferrin Absorption spectra of iron-loaded preparations of recombinant h T F / 2 N (0.36 mM) and h T F (0.40 mM) were recorded using a Cary 219 spectrophotometer ( all protein in 10 m M N a H C 0 3 ) . for h T F / 2 N (21) and human serum transferrin (23.5) compared favourably. Titration of 3.68 A28O units of apoprotein with Fe n i (NTA )2 (Figure 33) showed a slope corresponding to an E465,mM = 2.1, yielding E280,mM = 38.8, both reasonable values for the transferrin half molecule (Lineback-Zins and Brew, 1980). U r e a - P A G E analysis of the four peak components (a-d) from F P L C is displayed in Figure 31, panel C. The apo forms of transferrin molecules run with greater mobility than iron-bound forms. The presence of both apo and holoprotein in these fractions is likely due to the limited resolution of the F P L C procedure and the presence of contaminating iron during the subsequent work up. In general, peaks a and c represent the apo form of the minor and major species respectively, while peaks b and d are the corresponding iron-bound forms. In comparison to the proteolytically-derived half molecules, the recombinant molecules showed better monodispersity. 4. KTF/2C Studies The production of a gene construction encoding the C lobe of hTF fused to the natural hTF signal sequence is illustrated in Figure 34. PCR amplification using the described oligonucleotides produced only a single band by agarose gel analysis whose size (1.1 kbp) agreed with the predicted length of the target product. Dideoxy sequence analysis of the termini of this product showed it to indeed code for the desired product, however when this sequence was cloned into the expression vector p N U T and introduced into B H K cells, immunoreactive product was not detected either by immunoprecipitation or radioimmunoassay. 106 0.25 10 m M F e ( I I I ) ( N T A ) 2 . A*L Figure 33: Iron Titration of the Apo form of Recombinant hTF/2N Fe i n(NTA)2 (10 mM) was used to titrate 3.68 A28O units of the major form of recombinant hTF/2N in 10 mM NaHCOs (1 mL volume). Visible spectra were recorded 5-10 minutes after each addition. To characterize the hTF/2C PCR product fully, sequence analysis was performed using the ExoIII deletion protocol (Heinikoff,1984). Two unintentional errors were found in the coding sequence, resulting in the replacement of Glu507 by Gly and Lys657 by Glu. Sequence analysis of the 5' splice site of the hTF/2C-pNUT construction showed that the initiator methionine had not been lost during the assembly of the vector. 107 Pstl L Met EcoRl dGdC Pstl EcoRl • dGdC 1—//-J SIGNAL N-DOMALN C-DOMAIN EcoRl 01igo2 EcoRl C-DOMAIN Smal SIGNAL Smal l I EcoRl 1 r r r m „ i Olieo 25 Cycles of PCR Smal Smal SIGNAL C-DOMAIN t 1. Smal 2. Clone into large Smal fragment ofpNUT 108 B amino acid 330 334 338 Wild-type ...Gly .Thr Cys Pro Glu Ala Pro Thr Asp Glu GGC ACA TGC CCA GAA GCC CCA ACA GAT GAA Oligo 1 ...GGG CTG TGT CTG GCT GCC CCA ACA GAT GAA hTF/2C ...Gly Leu Cys Leu Ala Ala Pro Thr Asp Glu I ....Signal | C-lobe.... amino acid -5 -1 |334 338 nucleotide 2140 2160 I J Wild-type ...AGG GCT GCC ACC AAG GTG AAG ATG.GGA ACG Oligo 2 AGG GCT GCC ACC CGG GTG AAG ATG GG (complement) Smal Figure 34: Assembly Strategy for Recombinant hTF/2C Coding Sequence Panel A: Two separate oligonucleotides were used as PCR primers to create the h T F / 2 C coding sequence. A n £ c o R I restriction fragment including the entire C lobe coding sequence was used as a template for 25 rounds of PCR amplification. Oliogonucleotide 1 includes a Smal recognition site and the natural hTF signal sequence at its 5'end and matches the coding sequence for amino acids 334 -341 of hTF at its 3' end. Oligonucleotide 2 matches sequence in the 3' nontranslated region of the hTF c D N A and introduces a second Smal recognition sequence at this site. Panel B: The nucleotide and amino acid substitutions effected by Oligonucleotides 1 and 2 are presented. Oligonucleotide 1 splices the natural signal sequence from the hTF c D N A to the C lobe coding sequence starting at position 334 which is positioned in the linker peptide domain which normally joins the two lobes of hTF. Oligonucleotide 2 introduces a Smal recognition site downstream of the normal translation termination site (nucleotides 2125-2127 of the hTF cDNA). 109 IV. Discussion A. Prokaryotic Expression Studies Bacterial expression systems have been used routinely in studies where the production of large quantities of recombinant protein is desired. As pioneered by Sligar and co-workers in their studies on cytochrome 65 (von Bodman et al., 1986) and myoglobin (Springer and Sligar,1987), the use of synthetic genes has been shown to constitute a powerful alternative to natural cDNA or gene sequences in heterologous expression systems. Many independent factors must be considered in the design and assembly of prokaryotic expression vectors. Typically these systems are based on high copy number vectors which incorporate strong, inducible transcriptional promoter elements such as P j a c of p U C vectors or the hybrid P t a c promoter (De Boer et al., 1983) of pKK223-3. The tight coupling of R N A transcription, translation and degradation in bacteria makes it difficult to quantify transcription rates precisely, however strong promoters such as those mentioned above likely provide high levels of transcription of even eukaryotic sequences. Rather, the achievement of high level protein expression hinges on the efficiency of sequence translation and the modification, folding, processing and targeting of the polypeptide chain. Message translation in prokaryotes begins with the binding of the 16S subunit of the 30S ribosomal subunit to consensus binding sites adjacent to initiator methionine codons (Shine and Delgarno,1975). Consensus sequences covering the ribosome binding site (RBS) and initiator methionine have been compiled (Scherer et al., 1980; Stormo et al., 1982) and have been summarized as follows: Rule 1: The typical sequence register is A G G N 6 - 9 A T G Rule 2: G should not occur at position -3 (see Figure 35) Rule 3: Less than two G residues should occur between positions -1 and -7. 110 Rule 4: Positions +5 and +10 should be either A or T A comparison of the sequence regions between the ribosome binding site and the initiator methionine codon for the five prokaryotic expression vectors tested is presented in Figure 35. The vector ApKK223-3 hTF/2N violates Rules 1 and 3; however, the same parent vector has been shown to drive expression of human prothrombin (D.R. Cleveland, University of British Columbia, personal communication). Our studies did not allow us to determine quantitatively whether the sequence near the translation start site actually affected the translation efficiency of the recombinant proteins, although the complete absence of antigenically active product from either A p K K 2 2 3 - 3 - h T F / 2 N or pUC18-hTF/2N suggests that transcripts from these constructions were translated with poor efficiency, or recombinant hTF/2N may simply be highly unstable and is thus quickly degraded. The translation of eukaryotic D N A sequences in bacteria could also be affected by the codon bias of the heterologous gene, though the significance of this factor cannot be readily evaluated. The complete lack of recombinant protein observed for the human erythrocytic cytochrome b$ expression plasmid is indicative of the sensitivity of bacterial systems to poorly resolved factors. The amino acid sequence of the core structure of human erythrocytic cytochrome b$ differs at only three positions compared to the bovine lipase-solubilized hepatic cytochrome £5, specifically asparagine 17, tyrosine 27 and methionine 70, though the carboxy and amino terminal regions differ substantially (see Figure 3). Although any of these substitutions could readily account for the inability of the protein to fail to incorporate heme, the complete absence of apoprotein in NaDodS04-PAGE analysis demonstrates that the protein is translated with poor efficiency. Studies by Gheysen et al. (1982) examining the expression of the SV40 t-antigen in a bacterial expression system have shown that translation efficiency can vary over 2 orders of magnitude depending upon the nucleotide sequences surrounding the I l l V e c t o r C o n s t r u c t i o n -7 -1+1 5 10 pUC19-bovLpb5 I I I I I AACAACA ATG AGC AAA GCT pUC19-humErb5 AGGA AACAACA ATG GCC GAA CAA AGC pUC19-b0VTpb5 AGGA ACCAACA ATG GCT CTC AAA TAC ApKIC223-3-hTF/2N AGGA AAACAGGGGCAT ATG AAA CAA AGC ACT pUC18-hTF/2N AGGA AACACAAT ATG AAA CAA AGC ACT Figure 35: Comparison of the Upstream Sequences from Various Prokaryotic Expression Constructions Sequences between the ribosome binding site (underlined) and the initiator methionine are shown for the bacterial expression constructions used in this study. The A residue of the initiator methionine is designated +1. ribosome binding site and initiator methionine. These differences are attributed to hypothetical R N A secondary structures which trap either the ribosome binding site, initiator codon or both in double-stranded regions and presumably affect the translation efficiency of such transcripts. The use of in vitro mutagenesis to randomize the sequence surrounding the initiator codon has been demonstrated to increase the translation efficiency of bacterial expression systems substantially. The use of overlapping oligonucleotides in the cytochrome 65 gene assembly process provides an economical alternative to the complete synthesis of both D N A strands. The ready availability of oligonucleotides of extended length (100 bases or more) has now made it feasible to assemble genes of considerable size, and the use of enzymatic second strand synthesis almost halves the material costs involved in gene synthesis. The region of overlap need only be sufficient to form a stable hybrid at 112 ambient temperature. The results of this study clearly demonstrate the need for thorough screening of the target sequences for potential illegitimate priming events. As illustrated by the present study, the fidelity of the target amino acid sequence is critical to the use of synthetic genes in the production of recombinant proteins. Unfortunately, the available amino acid sequence for bovine cytochrome b$ was reported from chemical sequencing studies which relied on extensive acid treatment of derivatized peptides, which often results in deamidation of the sample. Had we not had the authentic protein isolated from liver for comparison, there would not have been reason to suspect the behavior of the initial recombinant protein. Only the report of the full-length c D N A sequence for the protein eliminated the necessity of re-sequencing the entire hepatic cytochrome 65 protein. The incorporation of heme into the recombinant cytochrome molecules appears to be the limiting factor in determining overall yields of the holoprotein. The S64A substitution in the triple mutant background sequence provides one example of the limited ability of some mutants to incorporate heme. The holo form of this particular mutant could only be recovered following prolonged incubation with exogenous heme, even though NaDodS04-PAGE analysis showed the same total amount of protein as the high yield wild-type species. Interestingly, the S64A substitution in the corrected wild-type background could be isolated without the need for exogenous hemin, albeit total yields were reduced from that of the wild-type (60 mg for S64A versus 500 mg for w.t.). Similar low-yield mutants have been observed for mutants in which surface histidyl residues were substituted, even though these replacements were not predicted to affect the heme-binding crevice directly (M. Mauk, W. Funk, unpublished work). Nuclear magnetic resonance studies on the apoprotein form of rat liver cytochrome 65 have demonstrated the presence of a stable core structure which centers on a cluster of residues about Trp 22 (Moore and Lecomte,1990) . Presumably, the apoprotein must maintain its own structural stability prior to the insertion of heme in vivo (Shawver et 113 al., 1984). Consequently any substitutions that affect the stability of this conformation would be expected to limit the yields of recombinant protein. The rate limiting reaction in heme synthesis in virtually all organisms studied is 5-aminolevulenic acid ( A L A ) synthesis (Lascelles et al., 1964). Attempts at improving holoprotein yields by supplementing cultures with exogenous A L A were not successful. B. Analysis of the Q11,E13,D57 Triple-Mutant of Cytochrome b$ The incorrect sequence of the original recombinant cytochrome may be regarded as a serendipitous mutagenesis experiment. The altered electrochemical and conformational characteristics of the triple mutant have been substantiated by the solution of a 1.9A resolution crystal structure for the protein (Funk et al., 1990b) which shows only slight deviations from the authentic wild-type bovine cytochrome b$ model (Figure 4). The substitution of Asn57 by an aspartyl residue brings the 7-carbon of a negatively-charged carboxylate within 11.3A of the heme iron atom (Figure 36). As demonstrated by the midpoint potential determination of the N57D mutant, this replacement is solely responsible for the 17 mV decrease in reduction potential in the triple.mutant. The /9-carbon of Asp 57 lies within 3.3A of the methylene carbon atom of the heme 4-vinyl group. Presumably the carboxylate charge can stabilize the oxidized form of the protein through an electronic inductive effect via conjugation of the 4-vinyl group to the 7r-system of the porphyrin ring. This interpretation may however be oversimplified in that the dominant heme orientation in the triple mutant is present at only 60%, compared to 80% (Mathews, 1980) or 90% (Walker et al., 1988; Reid et al, 1987) reported for the authentic wild-type protein. As the dominant and alternate isomers display different reduction potentials (calculated as +0.8 mV and -26.2 mV respectively) (Walker et al, 1988), the observed drop in potential for the Asp57 114 Figure 36: Relative Positioning of Asp57 in the Crystal Structure of Recombinant Lpb$ In the crystal structure of the triple mutant form of recombinant Lpb$ (Ql 1,E13,D57), the /9-carbon of Asp 57 is located within 3.3A of the methylene carbon of heme vinyl 4 (dominant heme orientation). The introduction of a negative charge at this position decreases the reduction potential of the heme center by 17 mV, likely due to an inductive electronic effect transmitted to the iron center via heme vinyl 4. replacement may also reflect a difference in conformational preference of the heme. Regardless, the sensitivity of the protein to substitutions at position 57 has been confirmed by studies in which Met, Phe and Cys replacements have been introduced (P. Barker, W.D. Funk, A . G . Mauk, work in progress). Most dramatically, the heme prosthetic group of the N57C substituted protein undergoes a spontaneous conversion to form a chlorin derivative, almost certainly as a result of the participation of the cysteine thiol in the reduction of a heme pyrrole ring. 115 The replacements at positions 11 and 13 (Q and E respectively) are illustrated in Figure 37. These substitutions do not affect the net charge of the molecule, nor do they affect the reduction potential of the heme center. Perturbations of the hydrogen bonding pattern of this region of the molecule were observed in the crystallographic structure nonetheless. In the authentic wild-type protein, the Glu 11 carboxylate interacts with both Lys 14 and His 15, while substitution of glutamine at this position appears to eliminate such electrostatic interactions. In addition, the Gin 11 side chain forms hydrogen bonds to the side chain hydroxyl and amide nitrogen of Thr 8 which are not observed in the authentic protein. Although the N - and C-termini of the protein are not well ordered in the triple mutant crystal structure, the subtle structural differences at position 11 and 13 may affect the conformation at the termini and thus account for the lowered susceptibility of the triple mutant to the action of trypsin and for the altered F P L C elution characteristics of the Q11.E13 mutant following conversion to the trypsin-solubilized form. C . S64A Substitution and DME Replacement Studies The reduction potential of the heme center in a given cytochrome is governed primarily by the nature of the axial ligands. However cytochromes that employ identical ligand sets can vary by more than 300 mV, as seen in the c-type family (Lemberg and Barrett, 1973). Moore and Williams (1977) have summarized additional factors which govern the potentials of such heme centers as follows: 1. Dielectric constant of the heme pocket 2. Electrostatic charge on the axial ligand; higher negative charge yields a lower redox potential 3. 7r-acceptor power of the ligand; greater acceptor power yields a higher redox potential 4. Spin-state of the redox interconversion 116 Figure 37: Comparison of the Crystal Structures of Recombinant and Authentic Cytochrome b§ at Positions 11 and 13 In the triple mutant form of recombinant Lpb^ (Q11,E13,D57), the side chain of G l n l l forms hydrogen bonds with the side chain hydroxyl of Thr8 and a main chain amide. In the crystal structure of the authentic trypsin-solubilized form of bovine liver microsomal cytochrome 65 (Mathews et al., 1971b), the carboxylate oxygens of G l u l l form hydrogen bonds with the side chains of Lys 14 and His 15. 5. Steric factors favoring oxidized versus reduced form 6. Special electrostatic interactions In comparing the variants of recombinant cytochrome 65, only the first and last factors on this list would likely have bearing, provided of course that the substitution does not alter significantly the architecture of the protein. The dielectric constant of the heme pocket will vary according to the microenvironment of specific residues and must 117 therefore be considered to be heterogeneous. In general the heme crevice maintains a low dielectric constant and thus electrostatic interactions between the heme and protein can have a substantial affect on the reduction potential, as illustrated by the position 57 substitutions described in this study. The crystal structure of bovine hepatic cytochrome 65 clearly shows that heme propionate 7 forms hydrogen bonds with the both the side chain hydroxyl group and main chain amide of Ser 64 and causes the propionate to rotate towards the protein surface (Mathews et al., 1971b; Mathews et al., 1972) (see Figure 5). Analysis of the crystal structure of the reduced form of the protein, ferrocytochrome £5, demonstrated that reduction did not significantly displace any of the protein residues, though the binding of a cation by heme propionate 7 was observed in the reduced structure (Argos and Mathews, 1975). These authors suggested a model for the red-ox process in which the carboxylate of heme propionate 7 serves to stabilize the formal positive charge of the heme in the oxidized state. Upon reduction, the same carboxylate is freed to bind a cation from solution (see Figure 7). The work of Reid et al. (1984) in which substitution of dimethylesterified heme (DME) into the cytochrome raised the reduction potential by 64 mV is consistent with this model. As predicted by the model, removal of the negative charge on the propionates would de-stabilize the ferric form of the protein and consequently raise the reduction potential. Disruption of the hydrogen bonding pattern to heme propionate 7 would also be predicted to raise the reduction potential by destabilizing the carboxylate charge. However the potential of the S64A mutant was actually lower than that of the wild-type by several millivolts, indicating that this substitution slightly increased the stability of the ferric state. These results suggest that the hydrogen bond between heme propionate 7 and the main chain amide is sufficient to stabilize ferricytochrome 65 and that the removal of the second hydrogen bond by the alanine substitution may allow the carboxylate group to re-orient to a position in closer proximity to the heme center that increases Coulombic stabilization of 118 the F e m center. Attempts at crystallizing the S64A mutant protein are currently in progress to test this hypothesis. The esterification of propionate carboxyls in DME-substituted species results in an identical increase in reduction potential (67 mV) for both the recombinant wild-type and S64A cytochromes. These findings are consistent with the view that the regulation of heme centers by peripheral carboxylates is primarily a function of the carboxylate charge and its relative position with respect to the iron center. D. Expression of h T F / 2 N in Eukaryotic Cells The highly convoluted structure of transferrin and the large number of disulfide bridges in the molecule may have been major impediments to the expression of h T F / 2 N in bacteria. Our attempt to mimic partially the natural processing of transferrin by targeting the protein for bacterial membrane transport via an attached alkaline phosphatase signal sequence was not successful, forcing us to explore eukaryotic expression systems. The mutant dihydrofolate reductase activity of p N U T represents a significant advantage over other selection systems in that transfected cells can be immediately selected with high levels of M T X , abrogating the need for a dihydrofolate reductase deficient cell line or tedious selection amplification procedures. The relatively short generation time of B H K cells (approximately 12 hours, mid-log phase) and their general hardiness make them ideal for large-scale cultivation. Several attempts have been made to increase further the product yield of the p N U T - h T F / 2 N B H K cell line. Attempts to achieve "superinduction" of the metallothionein promoter by using low serum content medium in the presence of actinomycin D (Walden et al., 1987) proved unsuccessful, likely due to the decreased viability of the treated cells in low serum (results not shown). Clonal isolation of individual cells from the primary cell line has yielded a line producing almost twice the 119 level of h T F / 2 N as judged by radioimmunoassay while serial selection of this same overproducing line to 5 m M M T X failed to show any additional increase in product level ( B . K - C . Chow, A.B. Mason, R.C.Woodworth, unpublished results). Most significantly, the continued maintenance of roller cultures over an extended period (up to two months) causes a large increase in the levels of recombinant product, achieving a peak concentration of over 80 mg/L (R.C. Woodworth, A.B. Mason, unpublished results). Product levels of this magnitude allow us to recover sufficient product for experiments such as N M R spectroscopy and crystallization studies which require large amounts of protein. The recombinant product occurs in both a major and minor form which share identical N-termini and molecular size but differ in their relative elution times on Polyanion SI and in their mobilities in urea-PAGE. Combined, these results suggest the minor form does not result from proteolytic degradation of the recombinant protein, nor does it represent a species with an improperly cleaved signal peptide. The decreased mobility of the minor form on urea-PAGE gels suggests that it maintains a tighter conformation than either the major form or thermolysin-produced h T F / 2 N molecule. Recombinant h T F / 2 N recovered from cultures maintained with serum substitutes requires fewer chromatographic steps in the purification process, and the resulting product is composed of less than 1% of the minor species, suggesting strongly that the minor component is derived from the major form during the isolation process (R.C. Woodworth, personal communication). The spectroscopic and iron-binding behavior of the recombinant h T F / 2 N match closely those of diferric human serum transferrin and the thermolysin-derived h T F / 2 N (Lineback-Zins and Brew,1980; Zak et al., 1983). Subsequent characterization of the isoelectric point, carbohydrate status and X H nuclear magnetic resonance spectrum of the recombinant h T F / 2 N molecule are all in good agreement with results from the proteolytically-derived half molecule (R.C. Woodworth, A.B. Mason,unpublished results; 120 Funk et al., 1990a). E . hTF/2C Construction The complete absence of recombinant product from the h T F / 2 C - p N U T construction again precludes thorough analysis of the system although several factors could conceivably affect the efficiency of expression. This construction shares the same 5' untranslated region, initiator codon region and signal peptide sequence with h T F / 2 N -p N U T plasmid and thus at least the initiation of translation for the two constructions should, to a first approximation, be similar. The predicted amino acid sequence at the site of signal peptide cleavage for the recombinant hTF/2C compared favourably to consensus rules developed for such cleavage sites in eukaryotic proteins (von Heijne,1983), although the failure of this processing step would account for the absence of antigenically-recognized material. Alternatively, the de novo folding of the C -terminal lobe of hTF may require the presence of the N-terminal lobe which would be absent in the recombinant system. In support of this hypothesis, non-covalent interaction between N - and C-terminal half molecules of human serum T F have been shown to stabilize both lobes against denaturation by urea (Ikeda et al., 1985). The presence of two inadvertent substitutions in the nucleotide sequence of the h T F / 2 C construction were most likely introduced during the PCR amplification procedure as a result of replication errors by the Taq polymerase. This is a common observation in PCR amplification procedures and is attributed to the lower fidelity of the thermostable enzyme compared to typical E. coli D N A polymerase activities. Conceivably the observed substitutions (E507G and K657E corresponding to G A A - G G A and A A G - A G G codon substitutions respectively) could destabilize the protein and thus prevent the accumulation and detection of recombinant product, although neither of these substitutions involve conserved residues seen in other transferrin species (Anderson et al., 1989). The correction of these sequence errors by site-directed mutagenesis or by 121 re-assembly of other PCR-generated clones will ultimately resolve this possibility and is currently underway. As a second alternative, the expression of a full-length recombinant hTF molecule has now been accomplished (W.D: Funk, D . K . Banfield, R . T . A . MacGillivray, unpublished results). Conceivably, a convenient protease recognition sequence could be engineered within the linker peptide separating the two lobes of the molecule to allow the subsequent recovery of both half molecules. F. Additional Projects and Work in Progress The successful expression of recombinant forms of cytochrome b$ and human transferrin have led to the initiation of several projects designed to probe structural and functional properties of these two molecules. In one ongoing study, surface histidyl residues of bovine cytochrome £5 (residues 15, 26 and 80) have been substituted in order to produce molecules in which only a single surface histidyl residue remains. This leaves a unique target residue for chemical modifications such as the binding of Ru(NH3)6Cl3, and as such eliminates many of the side-reaction products, such as doubly and triply modified proteins, which complicate studies on the wild-type species. These replacements are also valuable for studying electron transfer from the heme iron to bound copper complexes which interact with surface histidyl residues (Reid et al., 1987). Recombinant cytochrome 65 has also been used in probing molecular interactions with cytochrome c (Burch et al., 1988; 1990). The availability of complete N M R signal assignments for both cytochrome 65 (Guiles et al.,1990) and cytochrome c (Wand et al.,1989), together with recombinant expression systems for both molecules (Funk et c/.,1990b-cytochrome 65: Pielak et a/.,1985-yeast iso-1 -cytochrome c), will allow manipulations of either or both binding partners. Completed substitutions at position 57 of cytochrome 65 include Met, Phe, and Cys (P.D. Barker, W.D. Funk, A . G . Mauk, unpublished results). The sensitivity of the electrochemical potential to substitutions at this position has been described in this 122 report and subsequently confirmed by these additional replacements. Initial characterization of the Cys57 variant introduces a thiol group which appears to be directly responsible for the conversion of protoporphyrin IX in the protein to a chlorin-type porphyrin, characterized by the presence of a reduced pyrrole ring. Studies are presently underway to define precisely the role of the thiol at position 57 in this reduction reaction, as initial results suggest the absence of a free thiol in the protein following conversion. The study of mutant forms of hTF/2N has been initiated with the completion and initial characterization of a D63S variant of the molecule (R.C. Woodworth, W.D. Funk, A .B . Mason, R . T . A . MacGillivray; unpublished results). This substitution mimics a ligand replacement at this site seen in the C lobe of melanoTF and was created to test the relative importance of this ligand in the binding of metals. In addition to directly coordinating the metal, Asp 63 also participates in a hydrogen bonding pattern which links this residue with Domain N i l of the molecule and thus may be essential in maintaining the closed conformation of holoTF (E.N. Baker, personal communication). Initial characterization of D63S-hTF/2N shows that it binds iron with a significantly decreased affinity versus wild-type and has a yellow-shifted absorption maximum characteristic of the yellow 20 kDa Domain II fragment from duck T F (Evans and Madden, 1984). Additional substitution studies currently in progress include a replacement of the same Asp 63 ligand by Cys with the goal of possibly altering the metal-binding specificity of the molecule, and the production of K206Q and E207H variants to examine the role of these residues in stabilizing the bound anion/metal complex as has been suggested from a comparison of the primary sequences of transferrin and lactoferrin (Anderson et al., 1989). 123 V . R E F E R E N C E S Aasa R., Malmstrom B.G. , Saltman P. and Vanngard T. (1963) Biochim. Biophys. Acta 75, 203-222. Abe K . , Kimura S., Kizawa R., Anan F . K . and Sugita Y. (1985) J. Biochem. (Tokyo) 97, 1659-1668. Adrian G.S., Korinek B.W., Bowman B.H. and Yang F. (1986) Gene 490, 167-172. Ainscough E.W., Brodie A . M . , MacLachlan S.J. and Ritchie V.S. (1983) J. Inorg. Biochem. 18, 103-109. Aisen P., Aasa R., Malmstrom B. and Vanngard T. (1967) J. Biol. Chem. 242, 2484-2493. Aisen P., Aasa R. and Redfield A . (1969) J. Biol. Chem. 244, 4628-4633. Aisen P. and Brown E.B. (1975a) Prog. Hematol. 9, 31-53. Aisen P. and Brown E.B. (1975b) Prog. Hematol. 14, 25-56. Aisen P., Koenig S.H., Schillinger W.E., Scheinberg D.H. , Mann K . G . and Fish W. (1970) Nature 226, 859-861. Aisen P. and Liebman A . (1972) Biochim. Biophys. Acta 257, 314-319. Aisen P. and Liebman A. (1973) Biochim. Biophys. Acta 304, 797-804. Aisen P. and Liebman A. (1978) In "Transport by Proteins" (G. Blanes, H . Sund, eds.) Walter de Gruyter, Berlin. Aisen P., Liebman A . , Pinkowitz R.A. and Pollock S. (1973) Biochemistry 12, 3679-3684. Aisen P., Liebman A . and Sia C . L . (1972) Biochemistry 11, 3461-3464. Aisen P. and Listowsky I. (1980) Ann. Rev. Biochem. 49, 357-393. Aldred A .R . , Howlett G.J . and Schreiber G . (1984) Biochim. Biophys. Res. Commun. 122, 960-965. Alderton G . , Word W.H. and Fevold H . L . (1946) Arch. Biochem. 11, 9-13. Alsaadi B . M . , Williams R.J.P. and Woodworth R.C. (1987) J. Inorg. Biochem. 15, 1-12. Anderson B.F. , Baker H . M . , Norris G . E . , Rice D.W. and Baker E . N . (1989) J. Mol. Biol. 209, 711-734. Anderson B.F. , Baker H . M . , Dodson E.J . , Norris G . E . , Rumball S.V., Waters J . M . and Baker E . N . (1987) Proc. Natl. Acad. Sci. U.S.A. 84, 1769-1773. Argos P. and Mathews F.S. (1975) J. Biol. Chem. 250, 747-757. 124 Arnold R.R. , Cole M . F . and McGhee J.R. (1977) Science, 1917, 263-266. Ashwell G . and Morell A . G . (1974) Adv. Enzymol. 41, 99-128. Atkinson T. and Smith M . (1989) In "Oligonucleotide Synthesis: A Practical Approach" (Gait M.J . , ed) IRL Oxford. Azari P.R. and Feeney R . E . (1958) J. Biol. Chem. 232, 293-299. Azari P.R. and Feeney R .E . (1961) Arch. Biochim. Biophys. 92, 44-52. Bach J.S., Dixon M . and Zerfas L . G . (1946) Biochem J. 40, 229-239. Bagby S., Barker P.D., Di Gleria H . , Hill A .O. and Lowe V.J . (1988) Biochem. Soc. Trans. 16, 958-959. Bailey S., Evans R.W., Garratt R .C. , Gorinsky B., Hasnain S., Horsburgh C , Jhoti H . , Lindley P.F., Mydin A . , Sarra R. and Watson J.L. (1988) Biochemistry 27, 5804-5812. Baker E . N . , Rumball S.V. and Anderson B.F. (1987) TIBS 12, 350-353. Baker E . N . , Anderson B.F., Baker H . M . , Haridas M . , Norris G . E . and Rumball S.V. (1989) X-ray Structural Studies on Human Lactoferrin in Both the Iron- and Apo-Forms. IXth International Conference on Proteins of Iron Transport and Storage, Brisbane, Australia. Bates G.W. and Schlabach M . R . C . (1975) J. Biol. Chem. 250, 2177-2181. Bezkorovainy A . (1966) Biochim. Biophys. Acta 127, 535-537. Bezkorovainy A . and Grohlich D. (1967) Biochim. Biophys. Acta 310, 365-375. Bezkorovainy A . , Rafelson M . E . and Likhite V. (1963) Arch. Biochim. Biophys. 103, 371-378. Bilimoria M . H . and Kamin H.C. (1973) Ann N.Y. Acad. Sci. 212, 428-448. Blumberg B.S. and Warren L . (1961) Biochim. Biophys. Acta 50, 90-96. Boulik M . D . (1961) Clin. Chim. Acta 6, 493-497. Bowman B.H. , Yang F. and Adrian G.S. (1988) Adv. Genetics 25, 1-36. Breathnach D., Benoist C , O'Hare K . , Gannon F. and Chambon P. (1978) Proc. Natl. Acad. Sci. U.S.A. 75, 4853-4860. Brock J .H. (1985a) Top. Mol. Struct. Biol. 7, 183-262. Brock J .H. (1985b) In Metalloproteins (Part 2) (Harrison P .M. , ed.) MacMillan, New York. Brock J .H. , Arzabe F.R., Richardson N.E. and Deverson E . V . (1978) Biochem J. 171, 73-82. 125 Brosius J . , and Holy A . (1984) Proc. Natl. Acad. Sci. U.S.A. 81, 6929-6936. Brown J.P., Hewick R.N. , Hellstrom I., Hellstrom K . E . , Doolittle R.F. and Drezer W.J. (1982) Nature 296, 171-173. Brown J.P., Nishiyama K . , Hellstrom K . E . and Hellstrom I. (1981) / . Immunol. 127, 539-546. Brown J.P., Wright P.W., Hart C . E . , Woodbury R . G . , Hellstrom K . E . and Hellstrom I. (1980) J. Biol. Chem. 255, 4980-4983. Brown-Mason A . and Woodworth R.C. (1984) / . Biol. Chem. 259, 1866-1873. Brunei F . , Ochoa A . , Schaeffer E . , Boissier F. , Guillou Y. , Cereghini S., Cohen G . N . and Zakin M . M . (1988) J. Biol. Chem. 263, 10180-10189. Burch A . M . , Rigby S.E.J., Funk W.D., MacGillivray R . T . A . , Mauk A . G . , Mauk M.R. and Moore G.R. (1988) Biochem. Soc. Trans. 16, 844-845. Burch A . M . , Rigby S.E.J,, Funk W.D., MacGillivray R . T . A . , Mauk M.R. , Mauk A . G . and Moore G.R. (1990) Science 247, 831-833. Cannon J.C. and Chasteen N.D. (1975) Biochemistry 14, 4573-4577. Casey J .L . , Hentze H.W., Koeller D . M . , Caughman S.W., Reuault T . A . , Klausner R.D. and Harford J.B. (1988) Science 240, 924-927. Chaconas G . and van de Sande J.H. (1980) Methods Enzymol. 65, 75-85. Chance B. and Williams G.R. (1954) J. Biol. Chem. 248, 4947-4955. Charlwood P.A. (1963) Biochem. J. 88, 394-398. Chasteen N.D. (1977) Coord. Chem. Rev. 22, 1-36. Chasteen N.D. (1983) Trend Biochem. Sci. (Pers. ed.) 8, 272-275. Ciehanover A . , Schwartz A . L . and Lodish H.F. (1983) Cell 32, 267-276. Cochet M . , Gannon F. , Hen R., Maroteaux E . , Perrin F. and Chambon P. (1979) Nature 282, 567-574. Cohen H.J. and Fridovich I. (1971) J. Biol. Chem. 246, 359-366. Cohn E .J . (1947) Experimentia 3, 125-136. Cohn E .J . , Strong L . E . , Hughes W.L., Mulford D.J., Ashworth J.N. , Melin M . and Taylor H X . (1946) J. Am. Chem. Soc. 68, 459-475. Cristiano R.J. and Steggles A.W. (1989) Nucleic Acids Res. 17, 799. Daruish N . , Fischer C.W. and Steggles A.W. (1988) Prot. Seg. Data Analysis 1, 351-353. 126 DeBoer H . A . , Comstock L . J . and Vasser M . (1983) Proc. Natl. Acad. Sci. U.S.A. 80, 21-25. Diamond A . , Devine J .M. and Cooper G . M . (1984) Science 225, 516-519. Doolittle R .F . (1984) Evolution of the Vertebrate Plasma Proteins. In "The Plasma Proteins: Structure, Function and Genetic Control" (F.W. Putnam, ed.) 2nd Ed. , Vol . 4, pp 317-359. Academic Press, New York. Doolittle R .F . and Feng Y. (1987) C.S.H. Symp. Quant. Biol. 52, 869-874. Dorland L . , Haverkamp J . , Schut B.L. , Vliegenthart J .F .G. , Spik G . , Strecker G . , Fournet B. and Montreuil J. (1977) FEBS Lett. 77, 15-20. Douglas R . H . and Hultquist D.E. (1978) Proc. Natl. Acad. Sci. U.S.A. 75, 3118-3122. Druyan R., DeBernhard B. and Robinowitz M . C . (1968) / . Biol. Chem. 244, 5874-5878. Dutton P.L. (1978) Methods Enzymol. 54, 411-435. Egami F . , Hahashi M . , Sato R. and Mori T. (1953) / . Biochem. (Tokyo) 40, 527-534. Ehrenberg A . and Laurell C- .B . (1955) Acta Chem. Sci. 9, 68-72. Enoch H . G . and Strittmatter P. (1979) / . Biol. Chem. 254, 6483-6488. Evans R.W. and Madden A . D . (1984) Biochem. Soc. Trans. 12, 661-662. Evans R.W. and Williams J.C. (1978) Biochem J. 173, 543-552. Evans R.W., Williams J. and Moreton K . (1982) Biochem J. 210, 19-26. Fleischer S.F., Fleischer B., Azzi A . and Chance B. (1971) Biochim. Biophys. Acta 225, 194-200. Fleming P.J., Dailey H.A. , Corcoran D. and Strittmatter P. (1978) J. Biol. Chem. 253, 5369-5372. Fontes G . and Thivolle L . (1925) Compt. Rend. Soc. Biol. 93, 687-689. Fukushima K . and Sato R.C. (1972) J. Biochem. (Tokyo) 71, 447-461. Funk W.D. and MacGillivray R . T . A . (1990) Molecular Biology of Transferrin In "Proteins of Iron Transport and Storage" (P. Ponka, R.C. Woodworth, eds.) C . R . C . Press Reviews, Boca Raton, Fl . Funk W.D., MacGillivray R . T . A . , Mason A.B. , Brown S.A. and Woodworth R .C . (1990a) Biochemistry 29, 1654-1660. Funk W.D., Lo T.P., Mauk M.R. , Brayer G.D. , MacGillivray R . T . A . and Mauk A . G . (1990b) Biochemistry 29, 5500-5508. Gaber B.P. and Aisen P. (1970) Biochim. Biophys. Acta 221, 228-233. 127 Gatermann K . B . , Rosenberg G . H . and Kaufer N.F. (1988) Biotechniques 6, 63-64. Galfand D . H . and White T.J. (1990) In "PCR Protocols: A Guide to Methods and Applications" (Innis M . A . , Gelfand D.H. and White T.J . , eds.) Academic Press, San Diego. George S.K., Najera L . , Sandoval R.P., Countryman C , Davis R.W. and Ihler G . M . (1989) J. Bacteriol 111, 4569-4576. Gheysen D. , Iserentent D., Derom C. and Fiers, W. (1982) Gene 17, 55-63. Giblett E.R. (1968) Bibl. Haematol. (Basel) 29, 10-20. Giblett E .R. , Hickman C G . and Smithies O. (1959) Nature 183, 1589-1560. Gilbert W. (1978) Nature 111, 50. Gitlow S.E. and Beyers M.R. (1952) J. Lab. Clin. Med. 39, 337-346. Goodfellow P.N., Banting G . , Sutherland R., Greaves M . , Soloman E . and Povey S. (1982) Somat. Cell Genet. 8, 197-193. Gorinsky B., Horsburgh C , Lindley P.F., Moss D.S., Parkar M . and Watson J .L. (1979) Nature 281, 157-158. Goubin G . , Goldman G.S., Luce J. , Weiman D.E. and Cooper G . M . (1983) Nature 302, 114-119. Graham I. and Williams J. (1975) Biochem J. 145, 263-279. Guiard B. and Lederer F. (1979) / . Molec. Biol. 135, 639-650. Guiles R.D. , Altman J. , Kuntz LD. and Waskell L . (1990) Biochemistry 29, 1276-1289. Harding C , Heuser J. and Stahl P. (1983) J. Cell Biol. 91, 329-335. Harris W.R. (1983) Biochemistry 22, 3920-3926. Hebbert D. and Morgan E . H . (1985) Blood 65, 758-764. Hegesh E . , Hegesh J. and Kaftory A . (1986) New England J. Med. 314, 757-761. Heimburger W., Heide K . , Haupt H . and Schulze H .E . (1964) Clin. Chim. Acta 10, 293-307. Heinikoff S. (1984) Gene 28, 351-359. Holloway P.W. and Katz J.T. (1972) Biochemistry 11, 3689-3696. Holmberg C G . and Laurell C - B . , (1945b) Acta Physiol. Scand. 10, 307-312. Holmberg C G . and Laurell C - B . , (1947) Acta Chem. Scand. 1, 944-950. Hopkins C.R. (1983) Cell 35, 321-328. 128 Hopkins L . L . and Schwartz K . (1964) Biochim. Biophys. Acta 90, 484-491. Hopkins L . L . and Trowbridge I.S. (1983) J. Cell Biol. 97, 508-511. Huntley T . E . and Strittmatter P. (1972) J. Biol. Chem. 247, 4641-4647. Iacopetta B.J. and Morgan E . H . (1983) / . Biol. Chem. 258, 9108-9114. Iacopetta B.J., Rothenberger S. and Kuhn L . C . (1988) Cell 54, 485-495. Ichikawa Y. and Yamano T. (1970) Biochim. Biophys. Res. Commun. 40, 297-305. Ikeda H . , Nabuchi Y . , Nakazato K . , Tanaka Y. and Satake K . (1985) FEBS Lett. 182, 305-310. Imai Y. and Sato R.C. (1977) Biochim. Biophys. Res. Commun. 75, 420-426. Inman J .K. , Coryell F .C . , McCall K . B . , Sgouris J.T. and Anderson H.D. (1961) Vox. Sang. 6, 34-52. Ito A . (1980) J. Biochem (Tokyo) 87, 63-71. Ito A . and Sato R. (1968) J. Biol. Chem. 243, 4922-4923. Jacq C. and Lederer F. (1974) Eur J. Biochem. 41, 311-318. Jamieson G . A . , Jett M . and DeBernardo S. (1971) J. Biol. Chem. 246, 3686-3694. Jansson I. and Schenkman J.B. (1977) Archs. Biochim. Biophys. 178, 89-107. Jeltsch J . M . and Chambon P. (1981) Eur. J. Biochem. 120, 291-295. Johansson B. (1958) Nature 181, 996-997. Johansson B. (1960) Acta Chem. Scand. 14, 510-512. Jarnum S. and Lassen N.A. (1961) Scand. J. Clin. Lab. Invest. 13, 357-359. Jolles J . , Mazurier J. , Metz-Boutigue M . H . , Spik G . , Montreuil J. and Jolles P. (1976) FEBS Lett. 69, 27-31. Jones K . A . , Kadonaga T . T . , Rosenfeld P.J., Kelly T.J. and Tijan R. (1983) Cell 48, 79-89. Kadlubar F .F . and Ziegler D . M . (1974) Arch. Biochim. Biophys. 169, 83-92. Kajihara T. and Hagihara B. (1968) J. Biochem (Tokyo) 63, 453-461. Karlsson K - A . , Pascher I., Samuelsson B.E. , Finne J. , Krusivs T. and Rauvala H . (1978) FEBS Lett. 94, 413-417. Keilin D. (1925) Proc. R. Soc. Lond. B98, 312-339. 129 Keilin D. and Hartree E .F . (1940) Proc. R. Soc. Lond. B129, 277-306. Keller R . M . and Wuthrich K . (1980) Biochim. Biophys Acta 621, 204-210. Keller W. and Pennell R.B. (1959) J. Lab Clin. Med. 53, 638-645. Keyes S.R., Alfano J.A. , Jannson I. and Cinti D .L . (1979) J. Biol. Chem. 254, 7778-7784. Kimura S., Abe K . and Sugita Y. (1984) FEBS Lett. 169, 143-146. Koechlin B.A. (1952) / . Am. Chem. Soc. 74, 2649-2653. Kuhn L . C . , Ponka P. and Schulman H . M . (1990) Iron-Transferrin Requirements and Transferrin Receptor Expression in Proliferating Cells In "Proteins of Iron Transport and Storage" (P. Ponka and R.C. Woodworth, eds.) C.R.C. Press Reviews, Boca Raton, Fl . Kuhn L . C . , McClelland A . and Ruddle F . H . (1984) Cell 37, 95-103. Kunkel T . A . (1985) Proc. Natl. Acad. Sci. U.S.A. 82, 488-492. Laemmli N . K . (1970) Nature 227, 680-685. LaMar G . N . , Budd D .L . , Vislio D.B., Smith K . M . and Langry K . (1978) Proc. Natl Acad. Sci. U.S.A. 75, 5755-5762. LaMar G . N . , Burns P.D., Jackson J.T., Smith K . M . , Langry K . C . and Strittmatter P. (1981) J. Biol. Chem. 256, 6075-6083. Lascelles J. (1964) Tetrapyrole biosynthesis and regulation. W.A. Benjamin, New York. Laurell C.B. (1960) Metal Binding Plasma Proteins In "The Plasma Proteins" (F.W. Putnam, ed.) p. 349. Academic Press, New York. Le K . H . D . and Lederer F. (1983) EM BO J. 2, 1909-1914. Lederer F . , Ghrir R., Guiard B., Cortial S. and Ito A. (1983) Eur. J. Biochem. 132, 95-102. Lee D.C. , McKnight G.S. and Palmiter R.D. (1978) / . Biol. Chem. 253, 3494-3503. Leger D. , Tordera V . , Spik G . , Dorland L . , Haverkamp J. and Vleigenthart J . F . G . (1978) FEBS Lett. 93, 255-259. Legrand D., Mazurier J. , Colavizza D., Montreuil J. and Spik G . (1990) Biochem. J. 266, 575-581. Legrand D. , Mazurier J . , Metz-Boutigue M . H . , Jolles J. , Jolles P., Montreuil J. and Spik G . (1984) Biochim. Biophys. Acta 787, 90-96. Lehninger A . L . (1970) Biochemistry, Worth Publishers, Inc., New York. Lemberg R. and Barrett J. (1973) "Cytochromes", Academic Press, New York. 130 Lestas A . N . (1976) Br. J. Haematol 32, 341-350. Leto T . L . , Roseman M.A. and Holloway P. (1980) Biochemistry 19, 1911-1916. Lineback-Zins J. and Brew K . (1980) / . Biol. Chem. 255, 708-714. Livingston D.J. , McLachlan S.J., LaMar G .N . and Brown W.D. (1985) J. Biol. Chem. 260, 15699-15706. L u A . Y . H . , West S.B., Vore M . , Ryan D. and Levin W. (1974) J. Biol. Chem. 249, 6701-6709. Lucero M . A . , Schaeffer E . , Cohen G.N . and Zakin M . M . (1986) Nucleic Acids Res. 14, 8692. Luk C - K . , (1971) Biochemistry 10, 2838-2843. Lum J.B., Infante A . J . , Makker D . M . , Yang F. and Bowman B.H. (1986) J. Clin. Invest. 77, 841-849. MacGillivray R . T . A . (1975) Ph.D. Dissertation, University of Miami. MacGillivray R . T . A . and Brew K . (1975) Science 190, 1306-1309. MacGillivray R . T . A . , Mendez E . and Brew K . (1977) In "Proteins of Iron Metabolism" (E.B. Brown, P. Aisen, J. Fielding, R.R. Crichton, eds.) Grove and Stratton, New York. MacGillivray R . T . A . , Mendez E . , Shewale J .G. , Sinha S.K., Lineback-Zins J. and Brew K . (1983) / . Biol. Chem. 258, 3543-3553. Makey D . G . and Seal U.S. (1976) Biochim. Biophys. Acta 453, 250-256. Maniatis F . , Fritsch E .F . and Sambrook J. (1982) "Molecular Cloning, a Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. Mann K . G . , Fish W., Cox A . and Tanford C. (1970) Biochemistry 9, 1348-1353. Markley J .L. (1975) Acc. Chem. Res. 8, 70-75. Marks C.B., Vasser M . , Ng P., Henzen W. and Anderson S. (1986) J. Biol. Chem. 261, 7115-7118. Marsh J.B. and James A . T . (1962) Biochim. Biophys. Acta 60, 320-328. Martin E . M . and Morton R . K . (1955) Nature 176, 113-114. Martin A.W., Huebers E . , Webb J. and Finch C.A. (1984) Blood 64, 1047-1052. Maruyama T. , Gojobor A.S. and Ikemura T. (1986) Nucleic Acids Res. 14, rl51-rl97. Masson P.L. and Heremans J.F. (1966) Protides Biol. Fluids 14, 115-121. Masson P.L. , Heremans J.F. and Schonne E . (1969) / . Exp. Med. 130, 643-649. 131 Masson P.L. , Carbonara A . O . and Heremans J.F. (1965) Biochim. Biophys. Acta 107, 485-500. Masson P.L. , Heremans J.F. and Dive C. (1966) Clin. Chem. Acta 14, 735-739. Mathews F.S. (1980) Biochim. Biophys. Acta 662, 375-379. Mathews F.S. (1985) Prog. Biophys. Molec. Biol. 45, 1-56. Mathews F.S. and Czerwinski E.W. (1976) Cytochrome b5 and Cytochrome b5 Reductase from a Chemical and X - R a y Diffraction Viewpoint In "The Enzymes of Biological Membranes", Vol. 4 (A. Martinosi ed.) Plenum Publishing, New York. Mathews F.S., Argos P. and Levine M . (1971a) Cold Spring Harbor Symp. Quant. Biol. 36, 387-394. Mathews F.S., Levine M . and Argos P. (1971b) Nature 223, 1519. Mathews F.S., Levine M . and Argos P. (1972) J. Mol. Biol. 64, 449-458. Mauk M.R. , Reid L.S. and Mauk A . G . (1982) Biochemistry 21, 1843-1851. McClelland A . , Kuhn L . C . and Rodde F . H . (1984) Cell 39, 267-275. McKnight G.S., Hammer K . E . , Kuenzel E . A . and Brinster R . L . (1983) Cell 34, 335-341. McKnight G.S., Lee D.C. , Hemmaplardh D., Finch C.A. and Palmiter R.D. (1980a) J. Biol. Chem. 255, 144-147. McKnight G.S., Lee D.C. and Palmiter R.D. (1980b) J. Biol. Chem. 255, 148-152. McKnight G.S. and Palmiter R.D. (1979) J. Biol. Chem. 254, 9050-9058. Meek J. and Adamson E.D. (1985) J. Embryol Exp. Morphol 86, 205-212. Messing J. (1983) Meth. Enzymol. 101, 20-28. Metz-Boutigue M . H . , Jolles J. , Mazurier J. , Schoentgen R., Legrade D., Spik G . , Montreuil J. and Jolles P. (1984) Eur. J. Biochem. 145, 659-676. Meyers E.W. and Miller W. (1988) Cabios 4, 11-17. Montreuil J . , Tonnelat J. and Mullet S. (1960) Biochim. Biophys. Acta. 45, 413-421. Moore C D . and Lecomte J.T.J. (1990) Biochemistry 29, 1984-1989. Moore G.R. and Williams R.J.P. (1977) FEBS Lett. 79, 229-232. Morgan E . H . (1983) Synthesis and Secretion of Transferrin in "Plasma Protein Secretions by the Liver" (Glaumann H . , Peters T. Jr. and Ledman C , eds.) Academic Press, London. Morgan E . H . , Marsagalia G . , Giblett E . K . and Finch C.A. (1967) J. Lab. Clin. Med. 69, 370-381. 132 Mullner E.W., Neupert B. and Kuhn L . C . (1989) Cell 58, 373-381. Nelson R . M . and Long G . L . (1989) Analyt. Biochem. 180, 147-151. Newman R., Scheider C , Sutherland R., Vodinelich C. and Greaves M . (1982) Trends Biochem. Sci. 8, 397-1982. Nisimoto Y . , Takevchi F. and Shibata Y. (1977) J. Biochem. (Tokyo) 82, 1257-1266. Oka T. and Schimke R.T. (1969) / . Biol. Chem. 43, 123-127, Okada Y . , Frey A . , Guenther T. , Vesch F. , Sabitini D.D. and Kreibich G . (1982) Eur. J. Biochem 122, 393-402. Okada Y. , Frey A . , Guenther T. , Vesch F. , Sabitini D.D. and Kreibich G . (1982) Eur. J. Biochem 122, 393-402. Omura T. , Siekevitz P. and Palade G . E . (1967) J. Biol. Chem. 242, 2389-2396. Oncley J .L . , Scatchard G . and Brown A. (1947) J. Phys. Colloid Chem. 51, 184-198. Oshino N. (1982) Cytochrome b5 and its physiological significance In "Int. Encyclopedia of Pharm. and Ther." (Schenkman T. and Kupfer D. eds.), Pergamon Press. Oshino N. (1978) Pharmac. Ther. A. 2, 477-515. Ozols J. (1989) Biochim. Biophys. Acta 947, 121-130. Ozols J. and Heinemann F.S. (1977) / . Biol. Chem. 252, 8549-8553. Ozols J. and Heinemann F.S. (1982) Biochim. Biophys. Acta 704, 163-173. Ozols J. and Strittmatter P. (1969) J. Biol. Chem. 244, 6617-6618. Palmiter R.D. , Norstedt G . , Gelings R . E . , Hammer R . E . and Brinster R . L . (1983) Science 222, 809-814. Palmiter R.D. , Behringer R.R., Quaife C.J. , Maxwell F . , Maxwell I.H. and Brinster R . L . (1987) Cell 50, 435-443. Palmour R . M . and Sutton H . E . (1971) Biochemistry 10, 4020-4032. Paltauf F. and Holasek A . (1973) J. Biol, Chem. 248, 1609-1615. Pappenheimer A . M . Jr. and Williams C M . (1954) J. Biol. Chem. 209, 915-929. Park I., Schaeffer E . , Sidoli A . , Baralle F . E . , Cohen G.N . and Zukin M . (1985) Proc. Natl. Acad. Sci. U.S.A. 82, 3149-3157. Parker W . C and Beam A . G . (1962) / . Exptl. Med. 115, 83-105. Passon P.G. , Reed D.W. and Hultquist D.E. (1972) Biochim. Biophys. Acta 275, 51-61. Phillips M . E . and Thorbecke G.J. (1966) Int. Arch. Allergy Appl. Immunol. 29, 553-557. 133 Pielak G.J . , Mauk A . G . and Smith M . (1985) Nature 313, 152-154. Pinkowitz R .A . and Aisen P. (1972) J. Biol. Chem. 237, 7830-7834. Pladziewicz J.R., Meyer T.J . , Broomhead J.A. and Taubelt I. (1973) Inorg. Chem. 12, 639. Plowman G.D. , Brown J.P., Enns. C .A. , Schroder J. , Nikinana B., Sussman H . H . and Hellstrom I. (1983) Nature 305, 70-72. Polis B.D. and Schmukler W. (1953) J. Biol. Chem. 201, 475-500. Poulik M . D . (1957) Nature 180, 1477-1479. Poulos T . L . and Mauk A . G . (1983) J. Biol. Chem. 258, 7369-7373. Price E . M . and Gibbons J. (1972) Biochim. Biophys. Res. Comm. 46, 646-651. Princiotto J.V. and Zapolski E.J . (1975) Nature 255, 87-88. Prochownik E . V . , Markham A . F . and Orkin S.H. (1983) J. Biol. Chem. 258, 8389-8394. Pugh E . L . and Kutes M . (1977) / . Biol. Chem. 252, 68-73. Putnam F.W. (1975) Transferrin In "The Plasma Proteins" (F.W. Putnam, eds.) Vol . 1, 2nd Ed. , Academic Press, New York. Rachubinski D. , Verman P.S. and Bergeron, J.J.M. (1980) / . Cell Biol. 84, 705-716. Rado T . A . , Wei X . and Benz, E.J . (1987) Blood 70, 989-993. Raw I. and Mahler, H.R. (1959) J. Biol. Chem 234, 1867-1873. Reddy V . V . R . , Kupfer D. and Caspi E . (1977) J. Biol. Chem. 252, 2797-2801. Reid L.S. (1984) Ph.D. Dissertation, University of British Columbia. Reid L.S., Mauk M.R. and Mauk A . G . (1984) J. Am. Chem. Soc. 106, 2182-2185. Reid L.S., Taniguchi V . T . , Gray H.B. and Mauk A . G . (1982) J. Am. Chem. Soc. 104, 7516-7519. Reid L.S., Gray H.B. , Dalvit C , Wright P.E. and Saltman P. (1987) Biochemistry 26, 7102-7107. Remade J . , Fowler S., Amar-Costesec A . and Berthet J. (1976) / . Cell Biol. 71, 551-564. Rochard E . , Legrand D., Mazurier J. , Montruil J. and Spik G . (1989) FEBS Lett. 255, 201-204. Rodgers K . K . , Pachapsky T .C . and Sligar S.G. (1988) Science 240, 1657-1659. Rogers J. (1985) Nature 315, 458-459. 134 Rogers M.J . and Strittmatter P.C. (1973) J. Biol. Chem. 248, 800-806. Rogers M.J . and Strittmatter P.C. (1975) J. Biol. Chem. 250, 5713-5718. Rogers T.B. , Feeney R . E . and Meares C.F. (1977) J. Biol. Chem. 252, 8108-8112. Rogers T.B. , Boerrsen T. and Feeney R . E . (1978) Biochem. J. 17, 1105-1109. Roop W.E. and Putnam F.W. (1967) J. Biol. Chem. 242, 2507-2513. Rose T . M . , Plowman G.D. , Teplow D.B., Dreyer W.J., Hellstrom K . E . and Brown J.P. (1986) Proc. Natl. Acad. Sci. U.S.A. 83, 1261-1265. Rosseneu-Moutreff M . Y . , Soetewey F. , Lamote K . and Peeters H . (1971) Biopolymers 10, 1039-1048. Rossman M . G . and Argos P.C. (1975) / . Biol. Chem. 250, 7525-7533. Rouault T . A . , Hentze M.W., Caughman S.W., Harford J.B. and Klausner R.D. (1988) Science 241, 1207-1211. Salemme F.R. (1976) J. Mol. Biol. 102, 563-568. Sanger F . , Nicklen S. and Koulson A.R. (1977) Proc. Natl. Acad. Sci. U.S.A. 74, 5463-5467. Schade A . L . and Caroline L . (1944) Science 100, 14-15. Schade A . L . and Caroline L . (1946) Science 104, 340-341. Schade A . L . , Reinhard R. and Levy H . (1949) Arch. Biochem. 20, 170-172. Schaeffer E . , Lucero M . A . , Jelsch J .M. , Py M . C . , Levin M.J . , Cohen G.W. and Zakin M . M . (1987) Gene 56, 109-116. Schafer D . A . and Hultquist D.E. (1983) Biochem. Biophys. Res. Commun. 115, 807-813. Scherer G . R . E . , Walkinshaw M.D. , Arnot S and Moore D.J. (1980) Nucleic Acids Res. 8, 3893-3907. Schultze H . E . and Heremans J.F. (1960) Molecular Biology of Human Proteins, Elsevier, Amsterdam. Schultze H . E . and Schwick G . (1957) Behring Werk-Mitt 33, 11-17. Searle P.F., Stuart G.W. and Palmiter R.D. (1985) Mol. Cell. Biol. 5, 1480-1484. Shawver L . K . , Siedel S.L., Krieter P.A. and Shires T . K . (1984) Biochem. J. 217, 623-632. Shewale J .G . and Brew K . (1982) / . Biol. Chem. 257, 9406-9415. Shine J. and Delgarno L . (1975) Nature 254, 34-38. 135 Simonsen C . C . and Levinson A .D . (1983) Proc. Natl Acad. Sci. U.S.A. 80, 2495-2499. Slaughter S.R. and Hultquist D.E. (1979) J. Cell. Biol. 83, 231-239. Slaughter S.R., Williams C . H . Jr. and Hultquist D.E. (1982) Biochim. Biophys. Acta 705, 228-237. Spatz L . and Strittmatter P. (1971) Proc. Natl. Acad. Sci. U.S.A. 68, 1042-1046. Spatz L . and Strittmatter P. (1973) / . Biol. Chem. 248, 793-796. Springer B.A. and Sligar S.G. (1987) Proc. Natl. Acad. Sci. U.S.A. 84, 8961-8965. Spik G . , Bayard B., Fournet B., Strecker G . , Bouquelet S. and Montreuil J. (1975) FEBS Lett. 50, 296-299. Starkenstein S. and Harvalik Z . (1933) Arch. Exptl. Pathol. Pharmakol. Ill, 75-92. Stayton P.S., Poulos T . L . and Sligar S.G. (1989) Biochemistry 28, 8201-8205. Stellwagen E . (1978) Nature 275, 73-74. Stoffel W. (1961) Biochem. Biophys. Res. Commun. 6, 270-273. Stoneheurner E . , Williams J.B. and Millet F. (1979) Biochemistry 18, 5422-5427. Stormo G.D. , Schneider T.D. and Gold L . M . (1982) Nucleic Acids Res. 10, 2971-2995. Strickland D . K . , Hamilton J.W. and Hudson B.G. (1979) Biochemistry 18, 2549-2554. Strittmatter C F . and Ball E . G . (1954) J. Cell Comp. Physiol. 43, 57-78. Strittmatter C F . and Velick S.F. (1956) J. Biol. Chem. 221, 253-264. Strittmatter P. and Ozols J. (1966) / . Biol. Chem. 241, 4787-4792. Struck D . K . , Sivta P.B., Lane M.D. and Lennarz W.J. (1978) J. Biol. Chem. 253, 5332-5337. Sugiyama T. , Mik i N . and Yamuno T. (1979) Biochem. Biophys. Res. Commun. 28, 73-78. Surgenor D . M . , Koechlin B.A. and Strong E . C (1949) / . Clin Invest. 28, 73-78. Sutton M.R. and Brew K . (1974) FEBS Lett. 40, 146-148. Tabor S. and Richardson C.C.C. (1987) Proc. Natl. Acad. Sci. U.S.A. 84, 4767-4771. Takagaki Y . , Radharkrishman R., Wirtz K.W.A. and Khorana H . G . (1983) J. Biol. Chem. 258, 9136-9142. Tan A . T . and Woodworth R.C. (1969) Biochemistry 8, 3711-3718. Taylor J.W., Ott J. and Eckstein F. (1985) Nucleic Acids Res. 13, 8749-8764. 136 Teale F.W.J. (1959) Biochim. Biophys. Acta 35, 543. Teng C .T . , Pentecost B.T., Marshall A . , Bowman B.H. , Laltey P.A. and Naylor S.L. (1987) Somat. Cel Mol. Genet. 13, 689-695. Thorbecke G.J . , Liem H . , Knight S. Cox K . and Muller-Eberhard V . (1973) / . Clin. Invest. 52, 725-731. Tomimatsu Y. , Kint S. and Scherer J.R. (1973) Biochem. Biophys. Res. Commun. 54, 1067-1074. Ulmer D.B. (1969) Biochim. Biophys. Acta 181, 305-311. Valcour A . A . and Woodworth R.C. (1987) Biochemistry 26, 3115-3120. van Bockxmeer F . M . and Morgan E . H . (1979) Biochim. Biophys. Acta 584, 76-83. van Oost B.A. , Edgell C.J.S., Hay C.W. and MacGillivray R . T . A . (1986) Biochem. Cell Biol. 64, 699-705. Veitch N.C. , Concar D.W., Williams R.J.P., and Whitford P. (1988) FEBS Lett. 238, 49-55. von Bodman S.B., Schuler M . A . , Jollie D.R. and Sligar S.G. (1986) Proc. Natl. Acad. Sci. U.S.A. 83, 9443-9447. von Heijne (1983) J. Mol. Biol. 159, 537-541. Wainio W.W. (1982) Chemistry and Biology of Mammalian b-Type Cytochromes In "Int. Encyclopedia of Pharm. and Ther." (Schenkman T. and Kupfer D., eds.) Pergamon Press. Wainio W.W. (1978) Pharmacol. Ther. A. 2, 359-372. Walden W.E., Daniels-McQueen S., Smith L . L . and Thach R . E . (1987) Gene 61, 317-327. Walker F . A . , Emrick D., Rivera J.E. , Hanquet B.J. and Buttlaire D.H.C. (1988) J. Am. Chem. Soc. 110, 6234-6240. Wand A . J . , DiStefano D .L . , Feng Y. , Roder H . and Englander S.W. (1989) Biochemistry 28, 49-55. Warner R .C . and Weber I. (1953) J. Am. Chem. Soc. 75, 5094-5101. Watts C. (1985) J. Cell Biol. 100, 633-637. West S.B., Levin W., Ryan D., Vore M . and Lu A . Y . H . (1974) Biochem. Biophys. Res. Commun. 58, 516-522. Williams J. (1975) Biochem. J. 149, 237-244. Williams J. (1974) Biochem J. 141, 745-752. Williams J. (1962) Biochem. J. 83, 355. 137 Williams J . , Elleman T . C . , Kingston B., Wilkins A . G . and Kuhn K . A . (1982) Eur. J. Biochem. 122, 297-403. Woodbury R . G . , Brown J.P., Yeh M . U . , Hellstrom I. and Hellstrom K . E . (1980) Proc. Natl. Acad. Sci. U.S.A. 77, 2183-2187. Woodworth R . C . (1986) J. Inorg. Biochem. 28, 245-252. Woodworth R . C , Morallee K . G . and Williams R.J.P. (1970) Biochemistry 9, 839-842. Woodworth R . C , Virkaitis L . M . , Woodburg R . G . and Fava R. (1975) In "Proteins of Iron Storage and Transport in Biochemistry and Medicine" (R.R. Crichton ed.) North-Holland, Amsterdam. Yang F . , Lum J.B. and Bowman B.H. (1985) Molecular Genetics of Human Transferrin. In "Protides of the Biological Fluids" (H. Peeters, ed.) pp. 31-34, Pergamon, Oxford. Yang F . , Lum J.B., McGil l J.R., Moore C M . , Naylor S.L., Van Bragt P.H., Baldwin W.D. and Bownam B.H. (1984) Proc. Natl. Acad. Sci. U.S.A. 81, 2752-2756. Yeh Y . , Iwai S. and Feeney R .E . (1979) Biochemistry 18, 882-889. Yoo M . and Steggles A.W. (1988) Biochem. Biophys. Res. Commun. 156, 576-580. Yoo M . and Steggles A.W. (1989) Biochem. Biophys. Res. Commun. 163, 18-24. Yoshikawa H . (1951) J. Biochem. (Tokyo) 388, 1-5. Young S.P. and Aisen P. (1981) Hepatology 1, 114-118. Zak O. and Aisen P. (1985) Biochim. Biophys. Acta 829, 348-353. Zak O., Leibman A . and Aisen P. (1983) Biochim. Biophys. Acta 742, 490-495. Zhang H . and Somerville C. (1988) Arch. Biochem. Biophys. 264, 343-347. Zoller M.J . and Smith M . (1983) Meth. Enzymol. 100, 458-500. 138 Appendix I: Measurement of the Reduction Potential of Biological Molecules Most protein molecules do not exchange electrons with an unmodified electrode efficiently. In this report, two methods have been used to overcome this phenomenon and determine the reduction potentials of various recombinant cytochrome 65 species. A brief summary of the technical and theoretical aspects of spectroelectrochemical and cyclic voltammetric methods is given below. 1. Spectroelectrochemistry The transfer of electrons to an oxidized species can be described by the general reaction: kf Ox + ne" Red (1) kr The distribution of reduced and oxidized species in such a reaction can be manipulated by adjusting the applied voltage to the system and is governed by the standard reaction potential (E° ) which is expressed in terms of the Nernst equation: E= E° + RT In [0x3 (2) nF [Red] Typically, optically transparent thin layer electrodes (OTTLE) consist of a thin minigrid, working electrode which is mounted across the width of a small cuvette. This working electrode covers the majority of the optical cross section of the cell and is of sufficient porosity to be transparent to transmitted light (ca. 60%). Application of a potential to the cell from an external source establishes the potential of the solution and the position of the red/ox equilibrium of the solution. The ratio of oxidized to reduced protein is determined at an appropriate wavelength. Potential measurements can only be made relative to a reference potential. For the sake of experimental convenience, the Saturated Calomel Electrode (SCE) is a useful reference: 139 H g / H g 2 C l 2 / K C l (saturated, aq.) E= +0.244 V vs. SHE @ 2 5 ° C Thus measurements made against the SCE are conveniently converted to the Saturated Hydrogen Electrode (SHE) scale by adding the factor 244.4 mV (25°C) . Electron transfer between the working electrode and the protein is achieved by the inclusion of a soluble small molecule mediator to the protein solution. Ru(NH3)5Cl3 has been shown to undergo red/ox reactions with cytochrome 65 and gold surfaces and thus serves to shuttle electrons between the working electrode surface and the cytochrome (Reid et al., J. Am. Chem. Soc. 104, 7516 (1982)). However, care must be taken with such a system to ensure that the solution has attained true electrochemical equilibration between each change in applied potential and the subsequent absorption spectrum determination. Solutions of relatively high ionic strength accelerate equilibration. 2. Cyclic Voltammetry Cyclic voltammetry involves the detection of current during the linear scanning of applied potential between two limits. Several electronic effects are observed at the surface of an active electrode and must be considered when interpreting cyclic voltammograms. The overall rate of heterogeneous electron transfer will depend primarily on the rate of electron transfer at the surface and on the rate of bulk transfer of electroactive species to and from the electrode surface. As proteins do not interact directly with unmodified electrodes, the electrode surface is usually functionalized with a coating of an organic promoter (see Armstrong et al., Quart. Rev. Biop. 18, 261 (1986)). In the present study, a small thiol-containing peptide ( K C T C C A ) from proteolytically-digested metallothionein was used to modify a gold electrode as has been previously described (Bagby et al., Biochem. Soc. Trans. 16, 958 (1988)). 140 For samples in which the rate of electron transfer is fast compared to the rate of bulk transfer, the reaction rate will be limited by rate of mass transfer from the electrode surface. Systems with this characteristic are electrochemically reversible (Nernstian), and it is implicit that species at the electrode surface are always at an equilibrium position governed by the electrode potential. At the other limit, samples in which the electron transfer rate is slow compared to the bulk transfer rate are referred to as irreversible. The electrochemical behaviour of proteins at modified electrodes is often characterized by rates between these two limits. Such quasi-reversible systems display Nernstian behaviour at a low potential sweep rate, but become increasingly irreversible at higher sweep rates. As the potential is raised from a state of full oxidation, reduction occurs at the electrode surface and the resulting faradaic current increases. Initially, substrate reduction increases with the rising potential as seen in Figure IA. As the process continues, a layer of reduced product forms at the electrode surface which can only be relieved by the diffusion-controlled removal of product. A cathodic peak current ( i p c ) results from the competition between increasing potential and the developing accumulation layer. Upon sweep reversal, a corresponding anodic peak current ( i p a ) is observed for analogous reasons involving an increasing oxidizing potential competing with the developing accumulation layer of oxidized products. A n estimation of ip C and i p a can be made by extrapolating the background (or capacitive) current as shown in Figure IA and the corresponding peak potentials (Ep C and Ep a ) can be assigned. The half-wave potential (E1/2) is found midway between E p c and E p a and is analogous to the midpoint or reduction potential, provided the kinetics of electron transfer are identical for both the reduced and the oxidized species. A n estimation of the reversibility of the system can be made by examining the peak current ( i p c ) as a function of sweep rate (y). The Cottrell equation describes the current i , at time t after a potential step as: 141 i(t)= n F A C 0 b u l k ( 7 r D 0 a ) 1 / 2 K a t ) with a=(nFi4)/RT (3) where n is the number of electrons involved, F is Faraday's constant, A is the area of the electrode, CQ ^ u ^ is the bulk concentration of oxidized species and D 0 is the diffusional coefficient of the oxidized species. From (3) it is apparent that the current, i , is proportional to v1!2, provided the red/ox process is limited by planar diffusion. Thus for fully reversible processes, a plot of the peak current, ip C versus vll2, yields a constant. For quasi-reversible couples, such as most protein systems, the reversible limit is achieved at low sweep rates, but at higher rates (>0.2 V s - 1 ) , the rate of electron transfer becomes limiting as is the case for the cytochromes 65 studied here (Figure IB). 142. -300 0 +300 Epc Epa mV vs. S H E 143 1/2 V Figure I: Cyclic Voltamogram of the D57 variant of recombinant Lpb5 Panel A: Cyclic voltammetry was performed on a solution of D57 Lpb$ (100/xM) using a peptide-modified gold electrode as described previously (Bagby et al., 1988). The corresponding values for the peak currents ( i p c and i p a ) , peak potentials (Ep C and E p a ) and the peak separation ( A E p ) were calculated from the sweep at 50 mVs . Panel B: A plot of the peak current versus v1/2 shows the quasi-reversible characteristics of many protein couples. At low sweep rate, the relationship is linear, obeying the Cottrell relationship (Equation 3). At higher sweep rates, the rate of electron transfer becomes limiting, and linearity is lost. 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items