Search

The coding sequence for human serum transferrin was assembled from restriction fragments derived from a full-length cDNA clone isolated from a human liver cDNA library. The assembled clone was inserted into the expression vector pNUT and stably transfected into transformed baby hamster kidney (BHK) cells, leading to secretion of up to 125 mg/L recombinant protein into the tissue culture medium. As judged by mobility on NaDodSO4-PAGE, immunoreactivity, spectral properties (indicative of correct folding and iron binding), and the ability to bind to receptors on a human cell line, initial studies showed that the recombinant transferrin, is identical to three commercial human serum transferrin samples. Electrospray mass spectrometry (ESMS), anion-exchange chromatography, and urea gel analysis showed that the recombinant protein has an extremely complex carbohydrate pattern with 16 separate masses ranging from 78,833 to 80,802 daltons. Mutation of the two asparagine carbohydrate linkage sites to aspartic acid residues led to the expression and secretion of up to 25 mg/L nonglycosylated transferrin. ESMS, anion-exchange chromatography, and urea gel analysis showed a single molecular species that was consistent with the expected theoretical mass of 75,143 daltons. In equilibrium binding experiments, the nonglycosylated mutant bound to HeLa S3 cells with the same avidity and to the same extent as the glycosylated protein and the three commercial samples. These studies demonstrate conclusively that carbohydrate has no role in this function.

The N-terminal lobe of recombinant human serum transferrin (residues 1 to 337) has been crystallized in a form suitable for high-resolution three-dimensional X-ray crystallographic analyses. Crystals are of the orthorhombic space group P2(1)2(1)2(1), with unit cell dimensions of a = 44.9 A, b = 57.0 A and c = 135.9 A, and diffract to beyond 2 A resolution. Further studies show that isomorphous crystals of specifically designed mutants of this protein can also be grown. Structural studies of both recombinant and mutant protein forms will provide a basis for understanding the mechanism by which human serum transferrin functions.

Five site-directed mutants of the N-terminal half-molecule of human serum transferrin have been expressed in baby hamster kidney cells and purified to homogeneity. Expression levels and overall yields varied considerably from the wild-type protein, depending on the mutant in question. The mutants are D63S, D63C, G65R, K206Q, and H207E and are based on mutations observed in a variety of transferrins of known sequence. Their molecular masses, determined by electrospray mass spectrometry, agree with theory, except for the D63C mutant, which appears to be cysteinylated. All mutants bind iron but with varying affinities; qualitatively, in increasing order D63S approximately D63C approximately G65R much less than wild type less than or equal to H207E much less than K206Q. In general, reduction of formal negative charge within the binding cleft shifts the visible spectral maximum of the iron complex toward the blue and reduces the affinity for iron, and increasing the formal negative charge shifts the visible maximum toward the red and increases the affinity for iron. The K206Q mutant is exceptional inasmuch as its visible maximum shows a blue shift, but its affinity for iron is the greatest of all of the mutants studied. All mutants reported, in addition to the wild-type protein, exhibit very similar visible molar extinction coefficients for the iron complex and very similar changes in extinction coefficients at 240 nm on binding Fe(III) or Ga(III). These results suggest that in all cases the bound metal ion is coordinated by two tyrosyl side chains.

The cDNA sequence of rabbit liver transferrin has been determined. The largest cDNA was 2279 base pairs (bp) in size and encoded 694 amino acids consisting of a putative 19 amino acid signal peptide and 675 amino acids of plasma transferrin. The deduced amino acid sequence of rabbit liver transferrin shares 78.5% identity with human liver transferrin and 69.1% and 44.8% identity with porcine and Xenopus transferrins, respectively. At the amino acid level, vertebrate transferrins share 26.4% identity and 56.5% similarity. The most conserved regions correspond to the iron ligands and the anion binding region. Optimal alignment of transferrin sequences required the insertion of a number of gaps in the region corresponding to the N-lobe. In addition, the N-lobes of transferrins share less amino acid sequence similarity than the C-lobes.

Expression of the amino-terminal lobe of human serum transferrin secreted into the culture medium by transformed baby hamster kidney (BHK) cells has been increased from the levels reported originally of 10-15 micrograms/ml to 55-120 micrograms/ml. Use of the serum substitute, Ultraser G, has facilitated isolation of the recombinant protein, resulting in approximately 80% recovery of expressed hTF/2N from the culture medium. In the three experiments described, 300-750 mg of recombinant protein was collected over a period of 25 days from five expanded surface roller bottles each containing 200 ml of medium (seven to nine collections). The use of alginate beads to encapsulate the transformed BHK cells provided no advantage over normal culturing over 25 days. A lag in production resulting in 30% less recombinant protein over this time period was observed. The production and isolation procedures described are easily handled by one person. The system is amenable to incorporation of isotopically substituted amino acids useful in NMR studies.

A human liver cDNA library was screened with a synthetic oligonucleotide, complementary to the 5' region of human transferrin mRNA, as a hybridization probe. The full-length human cDNA clone isolated from this screen contained part of the 5' untranslated region, the complete coding region for the signal peptide and the two lobes of transferrin, the 3' untranslated region, and a poly(A) tail. By use of oligonucleotide-directed mutagenesis in vitro, two translational stop codons and a HindIII site were introduced after the codon for Asp-337. This fragment was inserted into two different expression vectors that were then introduced into Escherichia coli. As judged by NaDodSO4-polyacrylamide gel electrophoresis and Western blot analysis, however, recombinant hTF/2N was undetectable in bacteria transformed by these plasmids. Concurrently, we developed a plasmid vector for the expression of recombinant hTF/2N in eukaryotic cells. In this case, a DNA fragment coding for the natural signal sequence, the hTF/2N lobe, and the two stop codons was cloned into the expression vector pNUT, such that the expression of hTF/2N was controlled by the mouse metallothionein promoter and the human growth hormone termination sequences. Baby hamster kidney cells containing this hTF/2N-pNUT plasmid secreted up to 20 mg of recombinant hTF/2N per liter of tissue culture medium. Recombinant hTF/2N was purified from the medium by successive chromatography steps on DEAE-Sephacel, Sephadex G-75, and FPLC on Polyanion SI. The purified protein was characterized by NaDodSO4-PAGE, urea-PAGE, amino-terminal sequence analysis, UV-visible spectroscopy, iron-binding titration, and proton NMR.(ABSTRACT TRUNCATED AT 250 WORDS)