CA2103583C - Recombinant transferrins, transferrin half-molecules and mutants thereof - Google Patents

Abstract

Recombinant transferrin, transferrin half-molecules and mutant transferrins hav-ing altered metal-binding or other properties are described. The recombinant transferrin molecules are expressed in functionnal form by stable eukaryotic cell lines such as baby hamster kidney cells transformed with an ex-pression vector encoding the recombinant molecule. The recombinant transferrins can be used in metal chelation therapy to bind and clear excess toxic metals in patients suf-fering from metal overloads.

Description

WO 92/13550 PC'T/US92/00928 RECOMBINANT TRANSFERRINS TRANSFERRIN
HALF-MOLECULES AND MUTANTS THEREOF
It Background of the Invention The iron-binding pseudoglobulins collectively 05 called transferrins or siderophilins comprise a class I., of proteins with strikingly similar features. X-ray crystallographic analyses of human lactoferrin (Anderson, B.F. at a2, (1987) Prnc_ Nafl nnaA Cn;
T A$4:1769-1773) and rabbit serum transferrin (Bailey, S. gt a1= (1988) Biochemistry 27:5804-5812) reveal that these proteins consist of two similar lobes connected by a short bridging peptide and that each lobe contains two domains defining a deep cleft containing the binding site for a metal ion and a synergistic anion.
Chicken ovotransferrin gene has been expressed in transgenic mice (McKnight, G.S. at a1= (1983) Cell (Cambridge, MA) ,34:335-341) and a fusion protein of part of rat transferrin with galactosidase has been in E coli (Aldred, A. gt al= (1984) expressed Biochem, Biovhvs. Res. Commun. 122:960-965). Except for this fusion protein, attempts to express transferrin or portions of the molecule in prokaryotic 05 systems have been unsuccessful (Aldred, A. Pt a1=
(1984) Biochem. Biophys. Res. Commun. 122:960-965).
The highly convoluted structure of the protein and large number of disulfide bridges in the molecule are probably the major impediments to expression in bacterial hosts. Attempts to mimic partially the natural protein folding environment by targeting the protein for bacterial membrane transport via an attached alkaline phosphatase signal sequence have been unsuccessful.

Summary of the Invention This invention pertains to recombinant transferrin, to recombinant transferrin half-molecules comprising at least the metal-binding domains of a single lobe (amino-terminal or carboxy-terminal) of transferrin and to stable cell culture system for expression of the transferrin. The recombinant transferrin can be expressed in stable, transformed eukaryotic cells, such as baby hamster kidney cells, to yield essentially homogeneous (monodisperse) preparations of the full or half-molecule forms. The invention also pertains to mutant transferrins and transferrin half-molecules which have metal-binding or other properties which are different from the natural (wild-type) form of the transferrin. These include mutant transferrins and transferrin half-molecules which bind iron or other metals more or less avidly than natural transferrin.

-3- 21p3 38 3 Transferrin half-molecules can be used in metal chelation therapy to treat individuals affected with abnormalities of metal regulation or with metal poisoning. For example, transferrin half-molecules, 05 especially mutant forms which bind iron with a higher avidity than natural transferrin, can be administered to iron-overloaded individuals, e.g., thalassemics, in order to clear excess toxic iron from their bodies.
In addition, half-molecules, or mutants thereof having altered metal ion selectivities, could be used to clear other toxic metals, e.g., lead, mercury, cadmium, copper and zinc from the body.

Description of the Figures = Figure 1 shows construction of the hTF/2N
expression vector in pNUT. A 2.3-kb cDNA encoding human serum transferrin was isolated from a human liver cDNA library and a 1.5-kb PstI/HaI fragment containing the complete amino-terminal domain coding sequence was cloned into M13mp18. Double translational stop codons and a HindIII recognition sequence were introduced by site-directed mutagenesis, allowing the isolation of a BamHI/HindIII fragment which, when joined to a BamHI/HpaII fragment, encodes = the amino-terminal domain and signal sequence. This fragment was cloned into the eukaryotic expression vector pNUT, giving the vector pNUT-hTF/N2. In this plasmid, the transferrin cDNA is under the control of the metallothionein promoter (MT-1 pro) and the human growth hormone transcription termination signals (hGH3'); pNUT also contains the SV40 early promoter (SV40) driving expression of a resistant DHFR cDNA
(DHFR cDNA) using transcription termination signals = from human hepatitis B virus (HBV).

'r,. . - . y . .. . . - . .
~ . . . .,, . . .. . -.. . . . . , Figure ~~.

2 shows a Western blot of immuno-precipitates from various baby hamster kidney cell lines. Samples of cell lysates (a) and medium (b) from Zn-induced cell cultures were precipitated with 05 anti-hTF antiserum. Samples of the resuspended pellets were analyzed by NaDodSo4-PAGE, transferred to nitrocellulose and developed with anti-hTF antiserum followed by alkaline phosphatase conjugated anti-IgG.
The hGH-pNUT and hTF/N2-pNUT cell lines were selected in 500 p M MTX and all cell culture was performed in DMEM/10% fetal calf serum. Lane 1, BHK cells; lane 2, hGH-pNUT transfected BHK cells; lane 3, hTF/N2-pNUT
transfected BHK cells. The positions of molecular weight markers (x 10-3) are indicated to the right of the blot, the position of the additional protein band of Mr 37,000 is also indicated (<37) to the right of the blot.
Figure 3 shows the isolation and PAGE analysis of hTF/2N. (Panel A) FPLC isolations on a column of Polyanion SI of recombinant hTF/2N (upper trace) and proteolytically derived hTF/2N (lower trace). (Panel B) NaDodSO4-PAGE (5-12% gradient of acrylamide) of molecular weight standards (lane Mr) and 3}-g of each of peaks a-d from panel A. (Panel C) Urea-PAGE under nonreducing conditions of the FPLC peaks a-d (recombinant hTF/2N species) and peaks e-h (proteolytically derived hTF/2N species) from panel A. The positions of the apo-protein (apo) and iron-bound protein (Fe) are indicated. The conditions used for FPLC are given under Materials and Methods.
FPLC fractions were pooled as follows; peak a (fractions 23-27), peak b (28-31), peak c (32-38), peak d (39-45), peak e (28-31), peak f (32-36), peak g (38-44), and peak h (46-51).

Figure 4 shows titration of the major form recombinant hTF/2N with 10 mM Fe(III)(NTA)2. The amount of protein was 3.68 A280 units in 1.00 mL of 10 mM NaHCO3. Visible spectra were run 5-10 minutes 05 after each addition of iron to the magnetically stirred cuvette.
Figure 5 shows proton magnetic resonance spectra of recombinant hTF/2N. (a) Fourier transform spectrum with a line broadening of 2 Hz. (b) Convolution difference spectrum with a line broadening of 4 Hz and DC = 4.0, NS = 68 500. The protein sample was 8 mg in 0.1 mL of 0.1 M KC1 in 2H20.
Figure 6 shows the 19F nuclear magnetic resonance spectrum of m-F-Tyr recombinant hTF/2N. The figure shows a Fourier transformation with a line broadening of 10 Hz, NS = 30,000. The protein sample was 6 mg in 0.1 mL of 0.1 M KC1 in 2H20; the reference was 0.1 M
trifluoroacetic acid in 2H20.
Figure 7 shows two separate oligonucleotides used as PCR primers to create the hTF/2C coding sequence.
An EcoRI restriction fragment including coding sequence for the entire carboxy lobe was used as a template for 25 rounds of PCR amplification.
Oligonucleotide 1 includes a SmaI 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' untranslated region of the hTF cDNA
and introduces a second SmaI recognition sequence at this site.

. ~ . . ~.

Description of the Invention Detailed This invention provides for the production of recombinant transferrin, recombinant transferrin half-molecules and mutant forms of full-length 05 transferrin and transferrin half-molecules which have altered properties, such as improved metal-binding capability, compared to the natural transferrin molecules. Recombinant transferrins can be produced in large quantities and in substantially homogeneous (monodisperse) form. For example, recombinant half-molecules of human serum transferrin can be produced as an essentially homogeneous preparation substantially free of other humari serum proteins. In contrast, half-molecules prepared by proteolysis of the holo-protein are difficult to purify and, in fact, the carboxy-terminal half of human transferrin cannot be satisfactorily prepared by proteolytic means.
Recombinant techniques also allow the application of.
mutagenesis to design and produce new forms of transferrin.
In general, a recombinant transferrin of this invention is produced by transfecting a suitable host cell with a nucleic acid construct encoding the transferrin, culturing the transfected host cell under conditions appropriate for expression and recovering the recombinant transferrin expressed by the cell.
The amino acid sequences for five transferrins have been reported (Jeltsch, J.-M. and Chambon, P. (1982) EMr. J. Biochem. 122:291-295; MacGillivray, R.T.A. gt al. (1983) J. Biol. Chem. .258:3543-3553;
Metz-Boutigue, M.-H. Pt al. (1984) Eur. J. Biochem.
145:659-676; Rose, T.M. gt al. (1986) Proc. Natl.
Acad. Sci. USA B3:1261-1265; Baldwin, G.S. and -7- 6~~03a o~

Weinstock, J. (1988) Nucleic Acids Res=
16:8720-8730). The DNA sequence for human serum transferrin has been determined (Yang, F. gt a1=
(1984) PrQc. Natl. Acad. Sci. USA ~Ll:2752-2756).
05 Full-length DNA for production of recombinant transferrins or truncated DNA encoding either the amino-terminal or carboxy terminal lobe of transferrin or a portion thereof can be obtained from available sources or can be synthesized according to the known sequences by standard procedures. In order to provide for secretion of the recombinant transferrin into cell culture medium, DNA encoding a transferrin signal sequence (or other signal sequence suitable for the expression system) is positioned upstream of the transferrin encoding DNA.
Mutant forms of transferrin and transferrin half-molecules can be produced by standard techniques of site-directed mutagenesis. See Taylor gt a1=
(1985) Nucleic Acids Res. 12;8749-8764; Zoller, M.J.
and Smith, M. (1983) Meth. Enzvmol 100:458-500. In particular, mutagenesis can be used to produce mutant transferrins which have metal binding properties that are different from natural transferrin. For example, mutants capable of binding iron more avidly than natural transferrin can be produced. To produce such mutants metal-binding domains can be mutagenized to replace one or more amino acids involved in binding with different amino acids. In human serum transferrin, the amino acids which are ligands for metal chelation are shown below (the number beside the amino acid indicates the position of the amino acid residue in the primary sequence where the first valine of the mature protein is designated position 1) terminal lobe Carboxy terminal lobe Amino (amino acids 1-337) (amino acids 343-679) Aspartic acid 63 Aspartic acid 392 Tyrosine 95 Tyrosine 426 05 Tyrosine 188 Tyrosine 517 Histidine 249 Histidine 584 In other types of transferrin, the numbering is different, the ligands (amino acids) are the same.
Other regions of transferrin control binding and these too can be targeted for mutagenesis. These are usually positively charged amino acids such as lysine, histidine or arginine. For example, a mutant transferrin half-molecule which binds iron more avidly than natural transferrin can be produced by replacing the lysine residue at position 206 with glutamine ( AAG-CAG ) .
The transferrin-encoding DNA is cloned into a eukaryotic expression vector containing appropriate regulatory elements to direct expression of the DNA.
A preferred eukaryotic expression vector is the plasmid pNUT described by Palmiter, R.D. gt a1= (1987) Cell 50:435-443. This plasmid contains the metallothionein promoter which includes transcription of the transferrin encoding DNA in the presence of heavy metal and transcription termination signals of human growth hormone. In addition, pNUT contains dihydrofolate reductase gene under control of the SV40 early promoter with transcription termination signals from human hepatitis B virus to allow selection in cell culture. The gene encodes a mutant form of the enzyme which has a 270-fold lower affinity for the competitive inhibitor methotrexate. This allows for 10~.
-9- ~~~~~~~~ `,-, the immediate selection of transfected cells in very high concentrations (0.5 mM) of methotrexate and abrogates the need for a recipient cell line that is deficient in dihydrofolate reductase. pNUT also 05 contains pUC18 derived sequences which allows it to be amplified in L. coli to provide sufficient amounts of the plasmid for transfection of recipient cells.
The expression vector containing the DNA encoding the transferrin is incorporated into an appropriate host cell. The preferred host cell is a eukaryotic cell which can be transformed with the vector to yield a stable cell line which expresses a functionally active transferrin construct. A particularly useful cell is the baby hamster kidney cell. Baby hamster kidney cells can be transfected with a vector carrying the DNA construct encoding a transferrin (such as the pNUT plasmid) to provide a stable cell culture system which expresses and secretes a functionally active transferrin (full or half-molecule). These cells are well-suited for economical, large scale growth and can be obtained from readily available sources.
Standard techniques, such as calcium phosphate coprecipitation or electroporation can be used to transfect the eukaryotic host cell with the vector.
The cell is then cultured under conditions appropriate to induce expression of the transferrin. For example, baby hamster kidney cells transfected with the pNUT
vector are stimulated to express the transferrin construct in the presence of heavy metals. Baby hamster kidney cells are preferably cultured in the medium Dulbecco's Modified Eagle's medium-Ham's F-12 nutrient mixture with the serum substitute Ultraser GTM
(Gibco) at about 1%.

After an appropriate culture period, the expressed and secreted transferrin can be recovered from the culture medium. Standard purification procedures can be employed to yield a substantially 05. homogeneous preparation of the recombinant transferrin. In one embodiment, the transferrin in the culture medium is saturated with iron and then purified by anion exchange chromatography.
The recombinant transferrins of the invention can be used to chelate and clear iron or other toxic metals from the body. The customary approach to iron chelation in vivo has been to assess a wide variety of = naturally-occurring siderophores of microbial origin and synthetic iron chelators for their physiological effects, primarily the ability to bind and clear iron from the body. Many such compounds have been studied with varying abilities to clear iron and often with unacceptable side effects (Pitt, C.G. 2-t al, (1979) J.
Pharm. Exn. Therap. 208:12-18). As a result, the only iron chelator used for clearing excess iron from humans remains deferoxamine, a cyclic peptide from Streptomyces pilosis.
A preferred transferrin for iron chelation therapy is a mutant transferrin half-molecule which binds iron more avidly than natural transferrin. The use of a mutant half-molecule allows for more efficient chelation and removal of the metal. A
particularly preferred mutant half-molecule is K206Q, described in the Exemplification below, which contains a glutamine rather than a lysine at position 206.
= A transferrin half-molecule is advantageous because unlike the holo-proteins, it passes through the glomeruli of the kidney and is excreted in the WO 92/13550 PCT/US92/00928 {
= ~>r.'..
~.: =.:~~

urine, so that metal is not only chelated but also cleared from the body. Moreover, the single half-molecules do not bind to transferrin receptors on the membrane of tissue cells and therefore do not 05 deliver iron to these tissues. Further, half-molecules of human transferrin would probably be recognized as "self" by the human body and therefore would not elicit an immunological response.
In addition, mutant half-molecules can be designed to have altered metal ion selectivities. The chelators could be used to clear other toxic metals from the body, e.g., lead, mercury, cadmium, copper and zinc.
For chelation therapy, the recombinant transferrin is administered to a patient in amounts sufficient to chelate the metal and reduce circulating levels below toxic levels. Generally, it is administered in a physiologically acceptable vehicle, such as saline, by a parenteral route (typically intravenously).
Recombinant full-length human transferrin can be used in nonserum supplements for cell culture media.
Transferrin is required for iron uptake by growing cells. The use of recombinant transferrin avoids the risk of contamination (with, e.g., HIV or hepatitis virus) associated with transferrin purified from human serum.
The invention is illustrated further by the following exemplification:

EXEMPL I EICATI dN

1. Production of Recombinant Transferrin Half-Molecule Comprising the Amino-Termina3. Lobe.
~IATER IALS
05 T4 DNA ligase, DNA polymerase I(K2enow fragment) and T4 polynucleotide kinase were purchased from Pharmacia-PL Biochemicals. Restriction endonucleases were purchased from Pharmacia-PL Biochemicals and Bethesda Research Laboratories. Oligodeo=yribo-nucleotides were synthesized on an Applied Biosystems 380A DNA Synthesizer. Nitrocellulose filters were obtained from Schleicher and Schuell, 32P-labeled nucleotides from New England Nuclear, goat anti-human transferrin antiserum from the Sigma Chemical Company, formalin-fixed StaRhyloc~c ~u;j aureus cells from Bethesda Research Laboratories, the Protoblot*
immunoscreening detection system from Promega, the oligonucleotide-directed mutagenesis kit from Amersharn, Dulbecco's modified essential medium and fetal bovine serum from Gibco, and anti-human transferrin monoclonal antibody HTF-14 was from the Czechoslovakian Academy of Sciences. All other reagents were analytical grade or purer.

METHQDS
IsQiatign of- liuman SeruM Transfprrzn (hTF) cDNA.
A human liver cDNA library constructed in the E, expression vector pKT-218 (Prochownik, E.V. gt ~l.
(1983) J. Bio1LChgm. 75 :8389-8394) provided by Dr.
Stuart Orkin, (Harvard University) was screened using a synthetic oligonucleotide coding for the *Trade-mark amino-terminal eight amino acids of serum hTF as a hybridization probe. The oligonucleotide corresponded to nucleotides 88 to 111 of the hTF CDNA sequence reported by Yang, F. gt .~t ,(198A) Proc. Natl, ca .
05 Sci, USA n:2752-2756). The oligonucleotide was end-labeled with T4 polynucleotide kinase and 32P-ATP
(Chaconas, G. and van de Sande, J.H. (1980) Met ods Eny,ymQ ,1- 61: 75-85 ), and used to screen approximately 105 colonies. Restriction endonuclease mapping of positive clones and DNA sequence analysis were performed by using standard procedures with pUC19 and M13mp19 vectors, respectively (Maniatis, T. t& a),, (1982) Molecular Cloningy a Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY;
Messing, J. (1983) Me;hods EnzvMol- 1_Q1:20-78; Sanger, F. gt a],, (1977) proc, Na tl. gc a c. Sc i_ i1SA
_7_A:5463-5467) .
FXpression Vector an C rII Cultu-,re. The eukaryotic expression vector pNUT (Palmiter, R.D. gt Al, (1987) Cel1 (Cambridge, MA) U:435-443) and baby hamster kidney (BHK) cells were provided by Dr.
Richard D. Palmiter (Howard Hughes Medical Institute, University of Washington). After synthesis, oligonucleotides were purified on C18 reverse-phase columns (Sep-Pak, Waters Associates; Atkinson. T. and Smith, M. (1984) 01igo.~ucleotide Svnthesi,gl A
Practi,ra] Appr ach (Gait, M.J., Ed.) pp 35-81, IRL
Press, Oxford). Site-directed mutagenesis was performed by using the method of Taylor, J.W. gt Al, (1985) Nuc1ei c Ac,i.dsRgs; 11:8749-8764). Plastnid DNA
was prepared from Ec_Qli JM105 and purified by two successive centrifugation steps with cesium chloride density gradients.

*Trade-mark WO 92/13550 PC.T/US92/00928 BHK r~~03~53 cells were grown in Dulbecco's modified essential medium (DMEM) with 10% fetal bovine serum to approximately 107 cells per 10-cm dish and were subsequently transfected with l0pg of plasmid by the 05 calcium phosphate co-precipitation technique described by Searle, P.F. gt D_L, (1985) Mol. Cell. Biol.
5:1480-1489). After 24 hours, the medium was changed to DMEM containing 100 pM methotrexate (MTX) and surviving cells were serially selected to 500 uM MTX.
In some experiments, cells were selected immediately with 500 VM MTX. Large scale roller bottle cultures were initiated by seeding approximately 5 x 107 cells into each 850 cm2 roller bottle containing 100 mL of DMEM-MTX. Cultures were induced at 80% confluency by the addition of ZnSO4 to the medium to a final concentration of 0.08 mM. The medium was harvested 40 hours later.
Immune-precipitation and Western Blottina.
Immune-precipitation of cell culture medium and cell lysates was performed by the method of Van Oost, B.A.
-e-t al= (1986) Biochem. Cell Biol. 4:699-705).
Precipitates were analyzed by electrophoresis on 12%
polyacrylamide gels in the presence of NaDodSO4 (Laemmli, U.K. (1970) Nature (London) 227:680-685), followed by blotting onto a nitrocellulose membrane.
The blot was incubated in PBS containing 0.1 mg/ml gelatin, then treated with goat anti-hTF antiserum (250-fold dilution in PBS), and finally developed with an alkaline phosphatase-conjugated, rabbit anti-goat IgG antibody according to the supplier's instructions.

-1~-Am3no Agi ~SUbsti tntion. To incorporate 3-fluorotyrosine into the recombinant hTF/2N as a 19F
NMR probe, the culture medium was supplemented with D,L-m-fiuorotyrosine (Sigma Chemical Company) at 16%
05 of the concentration of L-tyrosine in the medium. The cells grew as well on this medium as on the medium lacking D,L-m-fluorotyrosine.
Iso a..ion of Recom inanthTF~. Harvested culture medium was made 0.01% in phenylmethylsulfonyl fluoride to inhibit proteases and sufficient Fe(III)(NTA)2 was added to saturate all transferrin in the medium. After stirring at room temperature, the solution was dialyzed for 24 hours versus cold running tap water, and then for a few hours versus Milli-Q*
purified water. Concentrated Tris-HC1 buffer, pH 8.4 was added to a final concentration of 5 mM, the preparation was centrifuged to remove any debris, and was loaded onto a column (2.5 z$0 cm) of DEAE-Sephacely'(Pharmacia) equilibrated with 10 mM
Tris-HCl buffer, pH 8.4.
The column was then eluted with a linear gradient, of NaCl (0 to 0.3 M) in the same buffer. Fractions showing a pink color were analyzed by NaDodSO4-PAGF, and fractions containing the recombinant protein (Mr 37,000) were pooled. Such fractions also contained bovine transferrin and albumin resulting from the fetal calf serum in the tissue culture medium. After concentration of the pooled fractions to 5 mL on an Amicon.YPM-10 membrane, the protein was chromatographed on a column (2.5 x 90 cm) of Sephader*G--75 Superfine*
(Pharmacia-PL Biochemicals) equilibrated with 100 mM
ammonium bicarbonate.

*Trade-mark Sometimes, a second chromatographic step through this column was necessary to resolve completely the hTF/2N from the bovine proteins. At this stage, the A465/A410 was usually < 1.0, indicating the presence 05 of a contaminating heme-protein (possibly hemopexin).
The hTF/2N was finally purified to homogeneity by FPLC
on a column (1 x 10 cm) of Polyanion SI*(Pharmacia) using a linear gradient of NaCI (0 to 0.3 M) in 50 mM
Tris-HC1, pH 8.0 over a period of an hour at a flow rate of lml/min. Fractions of 1 mL were collected.
Two to four protein bands emerged from the column, depending on the iron-binding status of the protein.
NaDodS04--PAGE was performed with 5% to 12$
gradient gels and urea-PAGE was performed according to a modification (Brown-Mason, A. and Woodworth, R.C.
(1984) J. g 1. r em. 2,U:1866-1873) of the Makey, D.G. and Seal, U.S. (1976) Biochim, B'on Ys. Acta 453:250--256 procedure. Electrofocusing was performed on a 0% to 50% sucrose gradient in a 110 mL glass column (LKB) with 0.8% Pharmalyte; pH 5 to 8 (Pharmacia). The column was prefocused overnight to a final current of 2 mA at 1000 V.
The protein sample in 0.2 mL was diluted with 5 mL of solution withdrawn from the middle of the gradient. The sample was then reinjected into the isodense region of the column and focusing was continued for 24 hours. The gradient was collected from the bottom of the column in 1.5 mL fractions.
Individual fractions were analyzed for A280 and for pH. Fractions with maximum A280 were selected as representing the pls of the apo- and iron-saturated proteins.

*Trade-mark iron was readily removed from the iron-protein by incubation in a buffer containing 1 mM NTA, 1 mM EDTA, 0.5 M sodium acetate, pH 4.9. The apo-protein was concentrated to a minimum volume on a Centricori*10 05 (Amicon), then diluted and reconcentrated twice with water and twice'with 0.1 N KCl. The apo-protein had a tendency to precipitate in pure water, but redissolved readily in 0.1 M KC1. The apo-protein was made 10 mM
in NaHC03 and titrated with a suitable concentration Xa of Fe(NTA)2 while monitoring the arisorbance at 465 nm.
OUantitative Immunoaggayv of Recornb nant hT,F/2N.
A competitive solid state immunoassay was used to assess the concentration of recombinant hTF/2N in the culture fluid and at various stages of the 15 purification (Foster, W.B. p& 31õ,, (1982) Thr=b, Res.
2$:649-661). Proteolytically-derived Fe-hTF/2N
(Lineback-Zins, J. and Brew, K. (1980) ]. B;o1. Chem, = :708-713) was radioiodinated (Fraker, P.J. and Speck, J.C., Jr. (1978) Diochgm. B Qphvs. Res. Commun.
20 J.Q:849-857) with Iodogen"-(Pierce Chemical Company) and used as the standard. The monoclonal anti-hTF
antibody HTF-14 was used as the probe (Bartek, J. pt Al, (1984) Folig Bio1. (Prague) U:137=140). This antibody recognizes only the amino-terminal lobe of 25 hTF (Mason, A.B. gt AI,_ (1988) Br. ,7__ Haematgl.
1:392-393) and does not recognize bovine transferrin (Penhallow, R.C. gt &J, (1956) J. r,e11. PhYSio =
12,1:251-260).
Amiino-terminal Secugnce Apalysi . The 30 amino-terminal sequences of both the minor and major-forms of recombinant hTF/2N were determined on an Applied Biosystems 470A Protein Sequencer in the Given Analytical Facility at the University of Vermont.

*Trade-mark WO 92/13550 PCl/U592/00928 Periodic Acid-Schiff Stain. The presence of oligosaccharides in the recombinant hTF/2N was determined by staining the protein with periodic acid-Schiff reagent (Fairbanks, G. gLt Al_, (1971) 05 Biochemistry U:2606-2617).
Nuclear Magnetic Resonance Spectroscopy. Proton and fluorine NMR spectra were obtained on the 5.872 Tesia Bruker WM NMR spectrometer in the Camille and Henry Dreyfus NMR Laboratory, Department of Chemistry, University of Vermont, operating in the Fourier transform mode with quadrature detection. An 19F
probe was provided by Dr. Christopher W. Allen of that department. For proton spectra, spectrometer settings were as described previously (Valcour, A.A. and Woodworth, R.C. (1987) ioc i~strv 2&:3120-3125).
For 19F spectra, the sweep width was 30,000 Hz, the acquisition time was 0.279 seconds, a receiver delay of 2.0 seconds intervened between acquisition and pulse of 15.0 is (900) and the sample was at 303 K.
19F chemical shifts are relative to O.lM
trifluoroacetic acid in 2H20. Protein samples were 6 to 8 mg in 0.1 mL of 99.8 atom% 2H20, and spectra were run on these samples in 0.1 mL capsules inserted into standard 5 mm NMR tubes containing 2H20. Free induction decays of 19F spectra were subjected to a line-broadening of 10 Hz prior to Fourier transformation.

RESULTS
Isolation of Human TF cDNA. Approximately '= 30 100,000 colonies of a human liver cDNA library (Prochownik, E.V. gt a1= (1983) J. Biol. Chem.
25$:8389-8394) were screened by using a 24 base -19- r ~a _ 0 3~~ ~n~
~~

oligonucleotide to the 5' sequence of the human TF
cDNA as a hybridization probe. A single positive colony was obtained. Extensive restriction enzyme mapping of the plasmid isolated from this clone agreed 05 completely with the patterns predicted from the human TF cDNA isolated from the same library by Yang, F. gt ,aly (1984) Proc. Natl. Acad. Sci. USA $U:2752-2756.
DNA sequence analysis of the 5'- and 3'-termini of this clone confirmed that it was identical to the full-length clone isolated by Yang et al. All subsequent sequence analysis performed during the mutagenesis and subcloning of this cDNA conformed exactly to the sequence reported previously.
Vector Construction and Expression. Two translational stop codons and a unique HindIil recognition site were introduced into the linker region between the amino- and carboxy-terminal domains of the hTF cDNA sequence by oligonucleotide-directed mutagenesis. The predicted translation sequence from this construct ends at Asp-337, according to the serum hTF numbering sequence (MacGillivray, R.T.A. p_t al, (1983) J. Biol. Chem. ZU:3543-3553).
The expression vector pNUT (Palmiter, R.D. gt l (1987) Cell (Cambridge, MA) .5Q:435-443) contains a mouse metallothionein-1/human growth hormone gene fusion that has been shown to direct high levels of human growth hormone in transgenic mice (Palmiter, R.D. gt al. (1983) Science (Washington, D.C.) 222:809-814). Important functional features of this vector include a mouse metallothionein-1 promoter to induce cDNA transcription in the presence of heavy metals, pUC18 sequences to allow replication and selection in E
coli, and a dihydrofolate reductase !'1 4 g35, 8 3) (DHFR) cDNA driven by the SV40 early promoter to allow selection in cell culture. The DHFR cDNA encodes a mutant form of the enzyme which has a 270-fold lower affinity for the competitive inhibitor methotrexate 05 (NM) (Simonsen, C.C. and Levinson, A.D. (1983) Proc.
Natl. Acad,Sci. y$g $Q:2495-2499). This allows for the immediate selection of transfected cells in very high concentrations (0.5 mM) of MTX and abrogates the need for a recipient cell line that is deficient in DHFR.
To construct the expression vector pNUT-hTF/2N, the BamHI-HindIII fragment from the bacterial expression vector was isolated (Figure 1). A
HpaII-BamHI fragment from the original transferrin cDNA clone was also isolated (Figure 1). These two fragments were then ligated into M13mp18 replicative form DNA that had been cut with Accl and HindiII.
Replicative form DNA from the resulting M13 phage was isolated, the insert released by cleavage with XbaI
and HindIII, and the ends made blunt ended. These steps ensured that the fragment included the translational stop signals, retained the natural signal sequence for the protein, and was free of the dG/dC tail found in the original vector (Figure 1).
This fragment was inserted into SmaI-cut pNUT, thus replacing the human growth hormone gene with a hTF/2N
encoding cDNA, but leaving the transcriptional termination signal from the growth hormone gene intact. This plasmid was transfected into BHK cells and the resulting transformants were selected in the presence of MTX.

(~ :~~. '=

C) To analyze the mRNA transcripts produced by the transfected BHK cells, total RNA was electrophoresed on an agarose gel in the presence of formaldehyde (Maniatis, T. g.t D_I, (1982) Mplecular Cloning, a 05 Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY). After transfer to nitrocellulose, the blot was analyzed by using an oligonucleotide to the 3' untranslated region of the hGH gene as a hybridization probe. An inducible mRNA of approximately 1.4 kb was detected in the transfected cell line but not in mock-infected BHK cells (data not shown). This agreed with the predicted size of the hTF/2N mRNA, including the expected hGH 3' untranslated sequence and poly (A) tail.
To analyze the polypeptides produced by the transformed BHK cells, Western blot analysis was performed both on cell lysates and the medium of various cell lines (Figure 2). Samples of BHK cells, BHK cells containing the hGH-pNUT plasmid, and BHK
cells containing the hTF/2N-pNUT plasmid were grown in DMEM (BHK cells) or DMEM-MTX (BHK cells containing pNUT vectors). When the cells were reaching confluence, samples of medium were taken and cell lysates were prepared. These samples were incubated successively with goat anti-hTF antiserum and formalin-fixed 5, aureus cells (Van Oost, B.A. gt a1=
(1986) Biochem. Cell Biol. _6_4:699-705).
Bound proteins were eluted by incubation with NaDodSO4, electrophoresed on a polyacrylamide gel, and transferred to a nitrocellulose membrane. The membrane was then incubated with goat anti-hTF
antiserum and rabbit anti-goat immunoglobulin conjugated to alkaline phosphatase. When cell lysates WO 92/13550 PCT/US92/00928 =

medium from BHK cells (Figure 2, lanes la and ib) or or BHK cells with hGH-pNUT plasmid (Figure 2, lanes 2a and 2b) were analyzed, only the expected goat immunoglobulin bands (Mr 25,000 and 50,000) from the 05 original goat anti-hTF antibodies and a small amount of cross-reacting material were observed. However, an additional band of Mr 37,000 was observed in cell lysates (Figure 2, lane 3a) or medium (Figure 2, lane 3b) of the BHK cells containing the hTF/2N-pNUT
plasmid. The molecular weight of this polypeptide chain is in excellent agreement with the molecular weight of the hTF/2N molecule (37,833) calculated from the amino acid sequence.
The homogeneity of the hTF/2N product indicates the successful removal of signal sequence as cell lysate and secreted samples comigrate on SDS-PAGE.
The anti-serum appears to be highly specific for human TF species, since little bovine TF is apparent in the precipitates.
In large scale cultures of the hTF/2N cell line grown in roller-bottles, the concentration of hTF/2N
in the medium was approximately 10-15 pg/ml as detected by radioimmunoassay.
Isolation andCharacterization of Recombinant hTF/2N. Recombinant hTF/2N was purified by a three-step procedure that led routinely to an 80%
yield of the major form of the protein, based on radioimmunoassay. The final purification on Polyanion SI led to quantitative resolution of the apo- and iron-saturated forms of both the minor (<5%) and major constituents of the protein (Figure 3, panel A), as corroborated by urea-PAGE (Figure 3, panel C). Note that on urea-PAGE the slowest moving bands are ?*'O 92/13550 PCI'/US92/00928 ~. .,.: , -23-~~~ ~35- 8 3 apo-hTF/2N and the faster moving bands are Fe-hTF/2N.
SDS-PAGE gels (Figure 3, panel B) showed the major and minor forms of recombinant hTF/2N to be monodisperse, of equal molecular weight and the major component to 05 be free of carbohydrate by PAS stain (data not shown).
In general these preparations appear to have better monodispersity than proteolytically derived hTF/2N (Lineback-Zins, J. and Brew, K. (1980) J. Biol.
CheID, M:708-713) (Figure 3). For example, the chromatographic peaks are more regular for the former, and the number of bands on urea-PAGE is greater for the latter. Spectral ratios for the iron-saturated recombinant protein are typically A280/A465 =21 and A465/A410 = 138, which compare favorably with values for pure diferric transferrin isolated from human plasma. Titration of 3.68 A280 units of the apo-protein with Fe(NTA)2 yields a slope corresponding to an E465MM = 2=1 and gives for the apo-protein E28OmM = 38.8 (Figure 4), both reasonable values for a half-transferrin molecule (Lineback-Zins, J. and Brew, R. (1980) J. Biol. Chem. 255:708-713; Zak, 0. Vt A2, (1983) Biochim. Bionhvs. Acta 742:490-495). The pI's for the apo- and Fe-hTF/2N were 6.5 and 5.4, respectively.
Amino-terminal sequence analysis of both the minor and major forms of recombinant hTF/2N gave results identical to those found (MacGillivray, R.T.A.
gt al. (1983) J. Biol. Chem. 2_$:3543-3553) for holo-hTF from serum (Table 1).

r'83 The proton NMR spectrum of the recombinant protein (Figure 5) is very similar to that for the proteolytically-derived hTF/2N (Valcour, A.A. and Woodworth, R.C. (1987) Biochemistry 2&:3120-3125), but 05 the resonance lines are sharper for the recombinant protein. The 19F NbII2 spectrum of the protein derived from a cell culture grown on medium supplemented with m-F-tyrosine (Figure 6) shows four well-resolved resonances, two possibly having an unresolved shoulder.

Table 1 Amino-Terminal Sequence of Human Transferrin and of the Recombinant Human Transferrin Amino-Terminal Half-Molecule Protein Amino Acid Sequence Reference human serum V-P-D-1C-T-V-R-W-C-A-V-S- MacGillivray transferrin (1983) recombinant V-P-D-K-T-V-R-W-X-A-V-S- this report hTF/2N (major) recombinant V-P-D-K-T-V- this report hTF/2N (minor) aThe recombinant hTF/2N sequences were determined on an Applied Biosystems 470A protein sequencer. Approximately 200 pmol of each sample was analyzed. bTwelve sequencer cycles were analyzed. cNo residue was identified at cycle 9; however, cysteine residues were not modified prior to the analysis. dSia sequencer cycles were analyzed.

-25- 91 09r_8J

By using recombinant DNA technology, a hTF/2N
molecule is produced that functions identically with the proteolytically derived species as judged by several independent criteria. This represents the 05 first reported expression in a stable cell culture system of a functionally active form of this important iron transport protein.
The pNUT based hTF/2N construction described here produces high levels of recombinant protein without the need for a DHFR-deficient cell line or tedious resistance amplification procedures. BHK cells are well-suited for economical, large scale growth and we are currently examining their growth characteristics on micro-carrier supports in bioreactor vessels. By using either roller bottles or a fermentor with a capacity of several liters, we can easily produce sufficient recombinant protein even for techniques such as NMR that traditionally have required a high concentration of protein.
The minor form of recombinant hTF/2N isolated on Polyanivn SI migrates more slowly than the major form on urea-PAGE (Figure 3, panel C), but at the same rate on SDS-PAGE (Figure 3, panel B). Thus, the apparent molecular weights are the same but the relative degrees of unfolding in 6 M urea differ. Note that the proteolytically-derived apo-hTF/2N shows even faster migrating species in 6 M urea (Figure 3, panel C, fractions g and h).
Contamination of apo-hTF/2N with Fe-hTF/2N and vice versa on these gels arises from the method of pooling FPLC fractions, from some loss of bound iron on the urea gel and from binding of contaminating iron during workup of the FPLC samples. Identical ~ ~.

,. , N-terminal sequences (Table 1) show that the signal peptide has been removed from both minor and major forms of the recombinant protein. As in hTF/2N from human serum (Lineback-Zins, J. and Brew, K. (1980) J.
05 Biol. Chem. M:708-713), the recombinant hTF/2N is non-glycosylated. The cause of the difference between major and minor forms of hTF/2N is unknown at present. The minor form has never represented more than 5% of the total recombinant protein and is usually less than 1%. Thus, the goal of isolating a monodisperse recombinant hTF/2N (the major form) has been achieved.
The iron binding behavior, pIs, migration on NaDodSO4-PAGE and urea-PAGE and proton NNII2 spectra of the recombinant hTF/2N match reasonably well those of the hTF/2N derived from amino terminal monoferric hTF
by proteolysis with thermolysin (Lineback-Zins, J. and Brew, K. (1980) J. Biol. Chem. 2,U:708-713; Valcour, A.A. and Woodworth, R.C. (1987) Biochemistrv 2&:3120-3125), except as noted above. The major form of the recombinant protein shows a higher degree of monodispersity (Figure 3) and its proton NNgt spectrum shows sharper resonance lines than does the proteolytically derived hTF/2N. There has been insufficient minor form for analysis by NMR.
Previous studies of the incorporation of m-fluorotyrosine into alkaline phosphatase from E.
coli have established the efficacy of 19F NMR for specifically probing the tyrosyl residues in a protein (Sykes, B.D. gt B-1, (1974) Proc. Natl. Acad S; USA
11:469-473; Hull, W.E. and Sykes, B.D. (1974) Biochemistry la:3431-3437). Incorporation of m-F-tyrosine into the recombinant hTF/2N proves that selective amino acid substitution is possible in this cell culture system and gives us access to a specific NMR probe of tyrosyl side chains. This preparation behaves in all respects like the non-modified protein 05 as described above for the non-substituted recombinant. When we have optimized the cell culture conditions to achieve higher levels of incorporation, changes in the 19F NNIl2 spectrum on addition of paramagnetic and diamagnetic metals and on changes in pH will be useful in studying the tyrosyl residues specifically involved in metal binding. Incorporation of selectively deuterated aromatic amino acids will allow us to dissect the aromatic region of the proton NMR spectrum of the protein in similar fashion to the studies on lysozyme from Japanese quail (Brown-Mason, A. gt Al, (1981) J. Biol. Chem. 251:1506-1509).

II. Production of Recombinant Transferrin Half-Molecule Comprising Carboxy Terminal Lobe.

An EcoRI restriction fragment including the coding sequence for the carboxy lobe of hTF was isolated from the full length hTF cDNA and then used as a templated for PCR-directed mutagenesis (Figure 2). Two oligonucleotides were synthesized to be used as PCR primers. Oligo 1 encodes a SmaI recognition site, followed by sequence encoding the natural signal sequence of hTF, followed by sequence matching the coding sequence for amino acids 334-341. The second oligonucleotide matches the complement of the 3' nontranslated region of the hTF cDNA and introduce a SmaI recognition sequence 3' to the normal translation termination site (nucleotides 2125-2127 using the numbering system of Yang, F. gt al. (1984) Proc. Natl.
Acad. Sci. USA $I:2752-2756). Twenty-five rounds of PCR amplification using Taq polymerase (Perkin Elmer) resulted in the desired DNA fragment which splices the 05 natural signal sequence of hTF to the C lobe coding sequence. This fragment was then digested with SmaI
and ligated with the large SmaI fragment of pNUT as for the hTF/2N expression studies.

III. Production of Recombinant Full Length Transferrin.
The coding sequence for human serum transferrin was assembled from restriction enzyme digestion fragments derived from the full-length cDNA clone isolated from a human liver library described above.
Since the parental plasmid (pKT-218) of the original clone had a limited number of unique restriction enzyme recognition sites, a series of cloning steps was required to introduce the coding sequence into a convenient vector. This process was initiated by cloning a HpaII/BamHI fragment from the 5' end of the cDNA into the vector pUC 18 (Messing, J. (1983) Meth.
Enzvmol= J~U:20-28). The resulting plasmid was digested with BamHI and HindIII and a BamHI/HindIIl fragment from the human transferrin cDNA was cloned adjacent to the initial fragment. The resulting plasmid was then digested with HindIII and PstI and a final HindIII/PstI fragment from the 3' end of the transferrin cDNA was cloned to complete the assembly of the full-length coding sequence. Digestion of the resulting plasmid with SacI and SphI released the full-length coding sequence as a single restriction fragment which was subsequently made blunt using T4 DNA polymerase and dNTPs and then cloned into the large Smai fragment of pNUT (Palmiter gl al.L (1987) 05 Ce11 U:435-443) as described for the N-and C-terminal transferrin half-molecule coding sequences.
Plasmid DNA was prepared from E.,_ CQli JM105 and purified by two successive centrifugation steps with cesium chloride gradients. Baby hamster kidney (BHK) cells were grown in Dulbecco's Modified Eagles' medium-Ham's F-12 nutrient mixture (DMEM-F-12) (Gibco;
Sigma) with 10% fetal bovine serum to approximately 107 cells per 100 mm dish and were subsequently transfected with 10 ul of plasrnid by the calcium phosphate coprecipitation technique described by Searle gt. ,a-L, (1985) M..pI, Cell Biol. E:1480-1489.
After 24 hours the medium was changed to pMEM-F-12 containing 500 ~M methotrexate to select the plasmid containing cells. Once selected, the cells were serially passaged at approximately 80% confluency with phosphate buffered saline containing EDTA (0.2 grn/1) to five 100-mm dishes, then to five T-175 flasks and finally to five expanded surface roller bottles (2o0 ml each). At the T-175 passage, a serum substitute, Ultraser G* (Gibco), at a level of 1% was used in place of fetal calf serum in DMEM-F-12 lacking phenol red.
It was found that once production levels were high (approximately 100 vg/ml of medium), medium without Ultraser G could sustain production of recombinant protein for at least two passages. This greatly simplified the isolation of the expressed full-length recombinant human serum transferrin. To isolate the recombinant protein, harvested culture *Trade-mark medium ~ ,. is made 0.01% with respect to phenylmethanesulfonyl fluoride and sodium azide to inhibit proteases and bacterial growth respectively.
Sufficient Fe3+ (nitrilotriacetic acid)2 is added to 05 saturate the transferrin present. The medium is reduced in volume to <10 ml and the transferrin is purified by passage over an anion exchange column (Polyanion SI, 1 x 10 cm) as described for the recombinant amino terminal human transferrin half-molecule. See above.
The isolated recombinant full-length human serum transferrin displays some heterogeneity on this column attributed to variation in the glycosylation pattern.
The protein is monodisperse on NaDod S04-polyacrylamide gel electrophoresis and has a spectrum and spectral ratios which are comparable to purified human serum transferrin.

IV. Production of Mutant Transferrins.
Substitution mutants are designated using the conventional single letter amino acid symbol of the wild type (native) residue, followed by the positional number of the replacement in the primary sequence, (where valine of the mature protein is designated position 1) followed by the symbol for the replacement residue. For example, a mutant in which aspartic acid residue at position 63 is replaced by a serine residue would be designated D63S.
The production of hTF/2N mutants was accomplished by two techniques. A D81S substitution was prepared using the method of Nelson, R.M. and Long, G.L. (1989) Analyt. Biochem. 180:147-151. Briefly, a HpaII/BamHI

WO 92/13550 PCT/US92/00928 }
..
-31- ~ ~ ,.~
... e~2;~'3c~
fragment from the 5' end of the hTF/2N coding sequence was subcloned into pUC18 and then used as a template for a two step PCR-based mutagenesis procedure. The resulting DNA fragment was then recloned into M13mp18 05 and the sequence of the mutant construction was confirmed by dideoxy sequence analysis. The fragment was then released from the double stranded form of the sequencing vector by digestion with XbaI and BamHI and then ligated to a BamHI/HindIII fragment from the original hTF/2N construction to produce a full length D81S-hTF/2N coding sequence, the fidelity of this splicing was confirmed by restriction digestion analysis and was subsequently cloned into pNUT as before.
The substitution mutants G65R, D63C, K206Q and H207E were produced by subcloning the entire hTF/2N
coding sequence into M13mp18, which was then used as a template for oligonucleotide-directed mutagenesis (Zoller, M.J. and Smith, M. (1983) Meth. Enzvmol.
1.QQ:458-500) using the dut-, ung- selection procedure (Kunkel, T.A. (1985) Proc. Natl. Acad. Sci. USA
12:488-492). Following mutagenesis, the entire coding sequence for the mutant sequences was confirmed by dideoxy sequence analysis using sequencing primers targeted along the length of the coding sequence at 250 bp intervals. The desired coding sequences were then released by restriction digestion, made blunt and inserted into pNUT as before.
pNUT plasmids have been constructed containing the cDNA a) for full-length human serum transferrin (hTF) and b) for various site-directed mutants of the amino-terminal half-molecule (hTF/2N). These mutants include 1) D63S patterned on the naturally occurring {
found in the C-terminal half of human mutation melanoferrin, b) G65R patterned on the naturally occurring mutant found in the C-terminal half of hTF
from a patient in England, c) K206Q based on the wild 05 tvpe mutation in the C-terminal half of ovotransferrin (oTF) from hen's egg white, d) H207E based on the wild tvpe mutation in human lactoferrin (hLTF) and e) D63C
as an attempt to change the metal selectivity of the iron binding site. All of these constructions have been expressed in stable transformants of baby hamster kidney cells in 10 to 100 mg amounts of recombinant protein. In addition pNUT plasmids have been constructed containing the full length cDNA for oTF
and chimeric cDNAs for hTF/2N-oTF/2C and oTF/2N-hTF/2C.
Characteristics of the site-directed mutants include: the D63S mutant does bind iron (contrary to speculations in the literature) but much less avidly than the wild type protein. For instance, this mutant loses its bound iron on electrophoresis in PAGE gels containing 8 M urea, whereas the wild type retains its bound iron. The maximum in the visible spectrum lies at 422 nm in contrast to that or the wild type at 470 rim. The G65R mutant binds iron less tightly than does the wild type and has a visible maximum at 470 nm.
The K206Q mutant binds iron much more avidly than does the wild type, as does its model, oTF/2C. Whereas the red color of the wild type iron protein disappears very rapidly in 0.5 M acetate buffer at pH 4.9, containing 1 mM each of EDTA and NTA, the mutant loses no color at all and requires pH 4 and 1 mM
deferoxamine to release its bound iron. The apo-mutant appears to rebind iron more slowly than the wild type protein. The visible maximum lies at 460 nm for this mutant.

The full length recombinant hTF runs at the same rate as the serum-derived protein on SDS-PAGE.
Equivalents Those skilled in the art will recognize, or be 05 able to ascertain using no more than routine experimentation, numerous equivalents to the specific procedures described herein. Such equivalents are considered to be within the scope of this invention and are covered by the following claims.

Claims (29)

CLAIMS:

1. A eukaryotic expression vector, comprising a nucleic acid construct comprising a nucleic acid encoding a full-length human serum transferrin or a human serum transferrin amino-terminal lobe linked to appropriate genetic regulatory elements for expression in an eukaryotic cell, wherein the nucleic acid construct comprises at least one mutation which results in an amino acid substitution, wherein the mutation does not inhibit expression of the encoded transferrin, the nucleic acid construct includes a nucleic acid encoding transferrin signal sequence linked to the nucleic acid encoding the full-length transferrin or transferrin amino-terminal lobe and the transferrin amino-terminal lobe contains a glutamine residue at position 206 in place of the lysine residue of natural transferrin.

2. A eukaryotic cell line transfected with the vector of claim 1.

3. A baby hamster kidney cell line transfected with the vector of claim 1.

4. A full length recombinant human serum transferrin mutant having a mutation at amino acid residue Asp63 or Gly65 wherein the mutant retains the ability to bind metal wherein Asp63 is mutated to serine or Gly65 is mutated to arginine.

5. The mutant of claim 4, wherein Asp63 is mutated to serine.

6. The mutant of claim 4, wherein Gly65 is mutated to arginine.

7. A recombinant human serum transferrin amino-terminal lobe mutant having a mutation at amino acid residue Asp63 or Gly65 wherein the mutant retains the ability to bind metal wherein Asp63 is mutated to serine or Gly65 is mutated to arginine.

8. The mutant of claim 7, wherein Asp63 is mutated to serine.

9. The mutant of claim 7, wherein Gly65 is mutated to arginine.

10. A full length recombinant human serum transferrin mutant having a mutation at Lys206 or His207 wherein the mutant has a stronger binding avidity for metal in comparison to the binding avidity of wild-type human serum transferrin wherein Lys206 is mutated to glutamine or His207 is mutated to glutamic acid.

11. A recombinant human serum transferrin amino-terminal lobe mutant having a mutation at Lys206 or His207, wherein the mutant has a stronger binding avidity for metal in comparison to the binding avidity of wild-type amino-terminal lobe of human serum transferrin wherein Lys206 is mutated to glutamine or His207 is mutated to glutamic acid.

12. The mutant of either claim 10 or 11, wherein Lys206 is mutated to glutamine.

13. The mutant of either claim 10 or 11, wherein His207 is mutated to glutamic acid.

14. A full length recombinant human serum transferrin mutant having a mutation at both Lys206 and His207, wherein the mutant has a stronger binding avidity for metal in comparison to the binding avidity of wild-type human serum transferrin wherein Lys206 is mutated to glutamine and His207 is mutated to glutamic acid.

15. A recombinant human serum transferrin amino-terminal lobe mutant having a mutation at both Lys206 and His207, wherein the mutant has a stronger binding avidity for metal in comparison to the binding avidity of wild-type amino-terminal lobe of human serum transferrin wherein Lys206 is mutated to glutamine and His207 is mutated to glutamic acid.

16. A eukaryotic expression vector comprising a nucleic acid encoding a full length human serum transferrin mutant having a mutation at either Lys206 or His207, wherein the mutant has a stronger binding avidity for metal in comparison to the binding avidity of wild-type human serum transferrin wherein the encoded mutant has Lys206 mutated to glutamine or His207 mutated to glutamic acid.

17. A eukaryotic expression vector comprising a nucleic acid encoding a recombinant human serum transferrin amino-terminal lobe mutant having a mutation at either Lys206 or His207, wherein the mutant has a stronger binding avidity for metal in comparison to the binding avidity of wild-type amino-terminal lobe of human serum transferrin wherein the encoded mutant has Lys206 mutated to glutamine or His207 mutated to glutamic acid.

18. The vector of either claim 16 or 17, wherein the encoded mutant has Lys206 mutated to glutamine.

19. The vector of either claim 16 or 17, wherein the encoded mutant has His207 mutated to glutamic acid.

20. A eukaryotic expression vector comprising a nucleic acid encoding a full length human serum transferrin mutant having a mutation at both Lys206 and His207, wherein the mutant has a stronger binding avidity for metal in comparison to the binding avidity of wild-type human serum transferrin wherein the encoded mutant has Lys206 mutated to glutamine and His207 mutated to glutamic acid.

21. A eukaryotic expression vector comprising a nucleic acid encoding a human serum transferrin amino-terminal lobe mutant having a mutation at both Lys206 and His207, wherein the mutant has a stronger binding avidity for metal in comparison to the binding avidity of wild-type amino-terminal lobe of human serum transferrin wherein the encoded mutant has Lys206 mutated to glutamine and His207 mutated to glutamic acid.

22. A eukaryotic expression vector comprising a full length recombinant human transferrin mutant according to any one of claims 4 to 6.

23. A eukaryotic expression vector comprising a recombinant human transferrin amino-terminal lobe mutant according to any one of claims 7 to 9.

24. A eukaryotic cell line transfected with the vector of any one of claims 16 to 23.

25. A baby hamster kidney cell line transfected with the vector of any one of claims 16 to 23.

26. A method of producing functionally active full length human transferrin, or mutant thereof, comprising:

a) constructing an expression vector comprising a nucleic acid encoding the full length human transferrin linked to appropriate genetic regulatory elements for expression of the full length human transferrin in a eukaryotic host cell;
b) transfecting the host cell with the expression vector;
c) culturing the transfected host cell under conditions conducive to expression of the full length human transferrin; and d) recovering the expressed full length human transferrin, wherein the full length human transferrin comprises at least one mutation consisting of Asp63 mutated to serine, Gly65 mutated to arginine, Lys206 mutated to glutamine, or His207 mutated to glutamic acid.

27. The method of claim 26, wherein the vector further comprises an inducible promoter of transferrin operably linked to the transferrin-encoding nucleic acid, said method further comprising the promoter in order to induce expression of transferrin.

28. The method of claim 27, wherein the promoter is the zinc inducible metallothionein promoter.

29. The method of claim 27, wherein the vector is the plasmid pNUT.