Physiological levels of binding and iron donation by complementary half-molecules of ovotransferrin to transferrin receptors of chick reticulocytes.

Two fragments, each corresponding to approximately half of the ovotransferrin (OTf) molecule and containing an iron-binding site were produced by digestion with affinity bound trypsin and were purified by isoelectric focusing and gel filtration chromatography. The immunologically distinct "half-molecules" individually have little ability to bind to transferrin receptors on chick embryo red blood cells or to donate iron to them. Combining them, however, leads to both binding and iron donation approaching that found for holo-OTf. Furthermore, similar amounts of radiolabeled iron can be extracted into the putative heme fraction from Fe2OTf and from the various combined half-molecules. These findings conflict with those reported by Keung and Azari ( (1982) J. Biol. Chem. 257, 1184-1188) for subtilisin-derived half-molecules of OTf examined in a similar system. They found that each half-molecule appeared to bind at a level of approximately one-third that of Fe2OTf and that the half-molecules competed with each other for binding sites. In contrast, our equilibrium binding studies, in the presence of 2,4-dinitrophenol to prevent iron removal, led to the determination of 4.79 X 10(4) binding sites/cell for Fe2OTf, 4.44 X 10(4) for the NH2-terminal half-molecules in the presence of excess COOH-terminal half-molecules and 4.17 X 10(4) for COOH-terminal half-molecules in the presence of NH2-terminal half-molecules; apparent binding constants were estimated to be 3.29 X 10(6), 1.19 X 10(6), and 0.67 X 10(6) M-1 for these same samples. Problems associated with equilibrium binding studies in which a narrow range of concentrations of ligand is used and/or iron is being removed are discussed. Labeled combined half-molecules were half as effective as labeled Fe2OTf in competition with unlabeled Fe2OTf. These findings are consistent with the lower apparent binding constant found in the equilibrium binding studies. Equimolar apo-OTf had no effect on binding of either Fe2OTf or the combined half-molecules. It seems apparent from our studies that the NH2- and COOH-terminal half-molecules each contain a recognition region both of which are necessary for binding to the transferrin receptor and iron donation to the chick embryo red blood cell.


Physiological Levels of Binding and Iron Donation by Complementary
Half-molecules of Ovotransferrin to Transferrin Receptors of Chick Reticulocytes* (Received for publication, May 9, 1983) Anne Brown-Mason, and Robert C. Woodworth Two fragments, each corresponding to approximately half of the ovotransferrin (OTf) molecule and containing an iron-binding site were produced by digestion with affinity bound trypsin and were purified by isoelectric focusing and gel filtration chromatography. The immunologically distinct "half-molecules" individually have little ability to bind to transferrin receptors on chick embryo red blood cells or to donate iron to them. Combining them, however, leads to both binding and iron donation approaching that found for holo-OTf. Furthermore, similar amounts of radiolabeled iron can be extracted into the putative heme fraction from FezOTf and from the various combined half-molecules. These findings conflict with those reported by Keung and Azari ((1982) J. Biol. Chem. 257, 1184-1 188) for subtilisin-derived half-molecules of OTf examined in a similar system. They found that each half-molecule appeared to bind at a level of approximately one-third that of FezOTf and that the halfmolecules competed with each other for binding sites.
In contrast, our equilibrium binding studies, in the presence of 2,4-dinitrophenol to prevent iron removal, led to the determination of 4.79 X lo4 binding sites/ cell for FezOTf, 4.44 X lo4 for the NHz-terminal halfmolecules in the presence of excess COOH-terminal half-molecules and 4.17 X lo4 for COOH-terminal halfmolecules in the presence of NHz-terminal half-molecules; apparent binding constants were estimated to be 3.29 x lo', 1.19 X lo', and 0.67 X 10' M" for these same samples. Problems associated with equilibrium binding studies in which a narrow range of concentrations of ligand is used and/or iron is being removed are discussed.
Labeled combined half-molecules were half as effective as labeled FezOTf in competition with unlabeled Fe20Tf. These findings are consistent with the lower apparent binding constant found in the equilibrium binding studies. Equimolar apo-OTf had no effect on binding of either FezOTf or the combined half-molecules.
It seems apparent from our studies that the NH2-and COOH-terminal half-molecules each contain a recognition region both of which are necessary for binding to the transferrin receptor and iron donation to the chick embryo red blood cell.

* This investigation was supported by United States Public Health
Service Grants F32-HL06231 and HL-23752. The multi-user y scintillation counter used for these studies was purchased with funds provided through the Research Advisory Council of the University of Vermont. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The transferrins (siderophilins) are a group of homologous glycosylated iron transport proteins which include serum transferrin from a variety of organisms, ovotransferrin from egg white, and lactoferrin from mammalian milk. Each protein molecule comprises a single polypeptide of about 80,000 Da, roughly divided into two similar domains, each containing a specific binding site for ferric iron and a concomitant obligate anion which physiologically may be carbonate or bicarbonate (1, 2). Transferrin binds to specific receptors on the surfaces of many different cells of animal origin (3) and is involved in iron transport to these cells.
Much research has been directed toward establishing the equivalence or nonequivalence of the two iron-binding sites. Confusion has come from the use of heterologous transferrinreticulocyte systems. It now appears that in isologous systems of rabbit, human, and chicken, the sites are functionally equivalent (4-6). Most chemical probes, however, suggest that physical nonequivalence of the two sites (7)(8)(9)(10)(11)(12).
One approach which has been taken with limited success is the use of proteolytically derived monoferric iron-binding fragments or half-molecules in an uptake system. Thus, it has been reported (13,14) that OTf' half-molecules are able to donate iron to rabbit reticulocytes, although no data have been presented to support the claim. In fact, it does not appear that even diferric OTf can function in this heterologous system (15). Brock et al. (16) report that rabbit reticulocytes take up little iron from bovine half-molecules. In preliminary studies with human NH2-terminal half-molecules, there is no binding to Tf receptors on human placenta or rabbit reticulocytes (17). As mentioned above, however, evaluation of halfmolecules derived from Tf of one species with reticulocytes from a different species is probably a poor experimental procedure. Furthermore, none of the studies cited reported the result of combining the half-molecules.
More recently, Keung and Azari (18) have published a report showing limited binding and iron-donating ability of subtilisin-derived half-molecules of ovotransferrin to chick embryo red blood cells. They found that binding of each halfmolecule was equal but the total number of binding sites at equilibrium was approximately one-third of the binding sites found for holo-OTf.
The NH2-terminal half-molecule appeared to be more efficient at donating iron to the cells than the COOH-terminal half-molecule but again both were con-siderably less efficient than Fe20Tf. Combining the halfmolecules led to no enhancement of binding; in fact, binding was lowered, perhaps showing that the two half-molecules were competing for the same site. In an experiment in which FezOTf competed against either of the half-molecules at equimolar concentrations, only 5% of the half-molecule remained bound. These results conflict, with our findings discussed below. As described herein, we have successfully made NH2and COOH-terminal half-molecules which when combined are active at levels comparable to FesOTf.

MATERIALS AND METHODS
Protein Preparation-Ovotransferrin was prepared as described previously (6). The NHz-terminal half-molecule designated FeOTf/ 2N was prepared from purified OTf, 45% saturated with iron nitrilotriacetate. A 0.5 X lo-, M Fe(NTA), solution, pH 7.1, was freshly made from a 0.50 M Fe(C104)3 stock solution before addition to the protein, which was 20 mM with respect to NaHC03. The monoferric species was isolated on an electrofocusing column (Servalytes, pH 4-9), dialyzed against glass-distilled HZO, and digested with DPCCtreated trypsin (Sigma) attached to Affi-Gel-10 (Bio-Rad). Both the sample and the Affi-Gel-trypsin were in 0.05 M Tris-HC1, 0.02 M CaClZ, 0.02% sodium azide, pH 7.8. After a 48-h incubation at room temperature on a Labquake (Fisher Scientific), the digested sample was reduced in volume and loaded onto a prefocused electrofocusing column (pH 4-9) and focused for 24 h. Final purification of the halfmolecule was achieved by chromatography on a Sephadex G-75 column (2.5 X 90 cm) equilibrated and run in 0.1 M NH4HC03.
The carboxyl-terminal half-molecule designated FeOTf/2C was prepared from FeOTf by dialysis against 0.5 M Na acetate, pH 5.5, until the ratio of A2so/A465 fell from 20 to about 50. One equivalent of ascorbate was added and the protein in a volume of 5 ml or less was passed over a Bio-Gel P-10 column (2.5 X 40 cm) in water. Fractions with an Az80/A,65 ratio of 50 were digested with Affi-Gel-trypsin and further purified as described above for FeOTf/2N.
Radiolabeling Procedures-After removal of iron from OTf or the half-molecules by dialysis against 0.2 M KPO,, 1 mM NTA, 1 mM EDTA, pH 6.0, then against glass-distilled water, apoprotein samples were made 20 mM in NaHC03. 59Fe-labeled protein was prepared by addition of radioiron (as carrier-free 59FeC13 in 0.1 N HCI, New England Nuclear) to an appropriate amount of a 20 mM stock solution of ferrous perchlorate in the presence of a 2-fold molar excess of NazNTA to give 95% saturation of the binding sites and a specific activity of 100 pCi/pmol of protein. Millimolar coefficients of absorbance at 280 nm of 91.9 mM-' for apo-OTf and of 114.0 mM" for Fe,OTf were used to determine the concentrations. Millimolar coefficients of half the above factors were used to estimate the concentration of each of the half-molecules. After incubation for 30-60 min at room temperature, samples were exhaustively dialyzed against 0.15 M NaCl, 20 mM NaHC03, 1 mM MgSO4 in the presence of Chelex-100 (Bio-Rad) and then lyophilized.
Iodination of Fe,OTf, FeOTf/2N, and FeOTf/2C was carried out by the iodine monochloride method of McFarlane (19) as described previously ( 6 ) . Specific activities of about 600 pCi/pmol of protein were sought. Nonprotein-bound iodine was removed by passing the sample over a desalting gel, Bio-Gel P-6DG (Bio-Rad), in an appropriate size syringe barrel fitted with a needle and a glass wool plug and disposed of after use. The effluent samples were exhaustively dialyzed against 0.01 M potassium phosphate, 0.15 M NaC1, 0.01% NaN3, pH 7.4, to assure the absence of free iodine. They were then dialyzed against glass-distilled HzO, lyophilized, and taken up in an appropriate volume of 0.15 M NaC1, 20 mM NaHCO,, 1 mM MgSO,.
Chick Embryo Red Blood Cells-Red blood cells were isolated from white leghorn chick embryos after 14-15 days of incubation at 37 "C and 80% relative humidity. The cells were prepared and incubations carried out (except as noted) as described by Williams and Woodworth (6). Cells were suspended in one of two media. The first, referred to as EGAB is a modified Earle's salt solution containing 0.12 M NaC1, 5 mM KCl, 1 mM NazPO,, 1 mM MgSO,, 1 mM CaClZ, 6 mM glucose, and 2.5 mg/ml of bovine serum albumin, made 20 mM in NaHC03 and equilibrated under 95% air, 5% CO, to pH 7.4 at 37 "C. The second, called EGA, differs from the above in that it is buffered to pH 7.4 by 20 mM Na,HPO, rather than by the COz/ bicarbonate system. Uptake Experiments-Washed, packed cells were diluted with 2 volumes of EGAB or EGA; 1.6-ml aliquots were pipetted into 7-ml polypropylene tubes (Sarstedt) fitted with rubber septa. After equilibration for 15 min under 5% C02,95% air in the case of EGAB and under air in the case of EGA, 20 nmol of radiolabeled protein was added in a volume of 0.4 ml of 150 mM NaC1,20 mM NaHC03,l mM MgSO,. Aliquots of 50 pl in triplicate were removed immediately and at various time points for each sample. Each aliquot was pipetted into 1 ml of ice-cold EGA in 1.5-ml conical tubes (Sarstedt), and centrifuged briefly in a Beckman Microfuge B. Following removal of the supernate by suction, two additional 1-ml washes were performed. The pelleted cells were counted in a Packard Prias 500C y counter. The numbers of OTf molecules bound/cell were determined as described below.
All uptake studies used about 10 p~ concentrations of proteins. Typically, the following 6 different samples were run: 1) 59Fe2'251- . The counts/min in each aliquot of cells was divided by a factor obtained by dividing the specific activity in counts/min/ nmol by 6.02 X lo', to obtain counts/min/molecule. This number was corrected for the number of cells in the aliquot as determined by hematocrit (6). Hematocrits were generally between 15 and 25% indicating 1.1-1.7 X lo9 cells/ml. Samples which contained both halfmolecules, but in which only one species was labeled, were considered only in terms of the specific activity of the labeled species. In other words, the nonradiolabeled iron and protein were not considered and the concentration of radiolabeled protein and iron is 10 p~. In terms of the total theoretical amount of 59Fe taken up per cell compared to 59FezOTf, a 50% level would be maximum. In a sample in which both half-molecules are labeled with iron and iodine, the iron specific activity is taken as the average of the specific activities of the two species, whereas the iodine specific activity is obtained by adding the specific activities of the two half-molecules together. It is obvious that, if the specific activities of the two half-molecules are very different, the Fe atoms/cell and sites/cell will not accurately reflect the relative contributions of each. Equilibrium Binding Studies-Washed cells were incubated in EGAB or EGA at 37 "C for 15-20 min and washed three times to eliminate as much endogenous OTf as possible. The cells were then incubated for 15-30 min at 37 "C in EGAB containing 0.75 mM 2,4dinitrophenol to inhibit iron uptake. After further washing, the cells were diluted 1:l with EGAB/DNP and 0.2-ml aliquots were pipetted into Omni-vials (VWR Scientific, Inc.) containing various concentrations of radiolabeled samples in a 0.1-ml volume. The samples were placed in a standard CO? incubator on a rocker platform for 30 min, at which time three 50-pl aliquots were removed and washed as described above. An aliquot of the supernatant from the hematocrit tube was assayed for radioactivity in order to determine the exact concentration of free protein in each vial.
The data shown in the figures for OTf and the half-molecules are for a given batch of cells on a given day, as some variability was found between cells on different days with respect to sites/cell and rates of iron uptake.
Preparation of Antibody-Antibody to OTf was raised in New Zealand White rabbits by established protocols (20). Pooled specific antisera dialyzed against phosphate-buffered saline were fractionated with 45% ammonium sulfate, dialyzed against phosphate-buffered saline, applied to an OTf affinity column (Affi-Gel-10, Bio-Rad), and eluted with HC1, pH 2.0, after extensive washing with phosphatebuffered saline. The specificity of the antibody was tested on Ouchterlony double diffusion plates (21).
Urea-Polyacrylamide Gel Ekctrophoresis-Electrophoresis in ureapolyacrylamide gels was carried out using methods derived from the original Makey and Seal protocol (22) as modified by Leibman and Aisen (23) and further by ourselves. Slab gels 10-cm long and 1. SDS-PAGE-SDS-PAGE was carried out in 5-12% acrylamide gradient gels (24), and stained by the method of Fairbanks et al. (25).

Binding and Iron Donation by Ovotransferrin Half-molecules
Molecular weight standards were obtained from Bio-Rad. Presentation of Data-Data from kinetic studies of binding of OTf and the half-molecules by CERBCs were fitted to a first order rate equation by the PAR biomedical computer program as previously described (26). This gives estimates of the first order rate constant, k, initial and final values of the dependent variable and their standard deviations. The lines in figures in which values of k are reported are the computer fitted lines to the experimental points presented in these figures.

RESULTS
The half-molecules are shown to be distinct by their immunological and physiological properties including Ouchterlony double diffusion plates (Fig. l), PI, molecular weight, and sequence at the NH, terminus ( Table I) as well as their NMR spectra (27). The sequence determination of the trypsinderived half-molecules is important in confirming the absence of significant internal nicking. SDS-gel electrophoresis (Fig.  2) reveals slight differences in molecular weight between the half-molecules and shows that disulfide bridges within each half-molecule appear to be intact since OTf and the halfmolecules migrate differently in the presence and absence of the reducing agent dithiothreitol. Furthermore, it should be noted that there is a small amount of holo-OTf in the final half-molecule preparations in addition to a band running below OTf/2N which is therefore of slightly lower molecular weight.
Urea-gel electrophoresis (Fig. 3) shows that the apo-and Fe-half-molecules migrate to positions distinct from apo-and Fe,OTf. The apo-forms of the two half-molecules can be   distinguished from each other whereas the iron-containing forms migrate to the same position in the gel.
After proteolysis with Affi-Gel-trypsin, the yield of FeOTf/ 2N is approximately 45% of theoretical, whereas the yield of FeOTf/2C is considerably lower and quite variable from preparation to preparation. The maximum yield achieved thus far is about 20% of theoretical. The purified iron-containing halfmolecules characteristically have A280/A465 ratios of about 20, the same ratio found for FezOTf.
Binding and Iron Uptake of 59Fe'251-labeled OTf and Halfmolecules by CERBCs-A typical binding profile and iron uptake is presented in Fig. 4, A and B. It is clear that the trypsin-derived half-molecules alone show little binding or ability to donate iron to CERBCs. Combining them, however, leads to binding and iron uptake approaching that of holo-OTf. Addition of the unlabeled complementary half-molecule, whether from the start of the experiment or at 30 min, leads to rapid binding of the labeled species. This type of experiment has been repeated a number of times with different batches of cells and different preparations of radiolabeled half-molecules. Table I1 presents results from four experiments in which the Fe atoms/cell and sites/cell found at 60 min for the various combinations of half-molecules are expressed as the per cent of Fe atoms/cell and sites/cell found for 5gFez1251-OTf. In the experiments presented, the sites/cell at 60 min for binding of 59Fe2'251-OTf were 1.74 X lo5, 6.49 X lo4, 6.31 X lo4, and 7.43 X lo4 and the Fe atoms/ cell were 5.33 X lo6, 3 Of possible significance is the observation that the Fe atoms/cell donated by 59Fe'251-OTf/2N to which FeOTfl2C was added after 30 min of incubation (Fig. 4B, sample 4) reaches only about half of the Fe atoms/cell after 60 additional min of incubation as are found for the same combination of samples incubated for 60 min from the beginning of the experiment (Fig. 4B, sample 3). In contrast, there is no significant difference between the Fe atoms/cell found after a 60-min incubation for 59Fe1251-OTf/2C whether the FeOTf/ 2N is present at the beginning of the experiment or is added after 30 min (Fig. 4B, samples 5 and 6). The experiment shown was done in EGA. In two subsequent experiments using the EGAB-C02 buffering system, the Fe atoms/cell after 60 min of incubation, whether from the start of the experiment or after the initial 30-min period, were identical for a given sample.
Competition of the Combined Half-molecules were Equimolar FezOTf and Apo-OTf- Table I11 presents the results of an experiment involving competition of the various combinations of half-molecules with equimolar amounts of either unlabeled FezOTf or apo-OTf. FezOTf at 10 p~ concentration reduced binding of labeled FezOTf by about half as would be expected. In the case of the various combined half-molecules, binding was reduced to about 25%. In other words, the labeled halfmolecules were half as effective as labeled FezOTf in competing with unlabeled FezOTf. Apo-OTf on the other hand had little effect on the binding of either labeled OTf (reduction of about 5%) or the various combinations of labeled half-molecules (reduction of about 15%).
The effect of equimolar FezOTf on labeled FezOTf was to reduce by half the Fe atoms/cell, again as expected. Apo-OTf had no effect on iron uptake from labeled OTf. For the various combined half-molecules, the presence of FezOTf reduced the Fe atoms/cell to about 15% of the control values, a slightly greater effect than was observed on the binding. Apo-OTf had little effect on Fe uptake from the combined half-molecules. Equilibrium Binding of OTf and the Half-molecules-Initial experiments with 59Fe'z51-OTf under aerobic conditions in EGA or EGAB showed that all of the radiolabeled iron was removed from the protein at concentrations below about 2 p~ during the 30-min incubation period. Since apo-OTf exhibits significantly different binding than FenOTf, the results are meaningless (see "Discussion"). Subsequently, 0.75 PM DNP was used to inhibit the removal of iron. It was found that a 30-min preincubation of cells with DNP was necessary to prevent iron uptake completely, as indicated by obtaining a ratio of 2 Fe atoms/site for 59Fe'251-OTf. The presence of glucose in the incubating medium did not overcome the effect of DNP. It must be noted that a significantly lower amount of binding was found with DNP-treated cells. In a dozen different uptake experiments in which cells were treated with DNP or not and incubated with 10 FM 59Fe'251-OTf for 30 min at 37 "C, there were an average of 24.1 4 5.8% fewer sites/cell for the DNP-treated cells than for the untreated cells.
A potential source of error, particularly in dealing with the half-molecules and DNP-treated cells, is the presence of endogenous OTf in the system. It was found that with 59Fe1251-OTf-loaded, DNP-treated cells that simple incubation at 37 "C in EGAB did not cause much OTf to be removed. Removal of labeled OTf was achieved after a chase with unlabeled FezOTf, indicating that binding is reversible. We therefore attempted to strip the cells prior to treatment with DNP. Although it has been reported (18, 28) that a 15-min incubation at 37 "C in EGAB removes endogenous OTf, our results in which cells were loaded with labeled Fe20Tf and then incubated a t 37 "C for different periods of time indicate that about 40% of the OTf remains associated with the cells after a 15-min incubation. In contrast, incubation in EGA for 15 min removes all but 13% of the original OTf. If the cells are then reincubated with labeled FepOTf, the sites/cell found in the original "uptake" are found.
In view of these considerations, equilibrium binding studies were conducted using stripped cells treated with 0. yields a long flat curve which bends upward at the higher concentrations (not shown). Fig. 6, A and B shows the results of plotting the data, uncorrected for nonspecific binding, according to the method of Scatchard (29) for the two sets of half-molecules. Fig. 7 presents the Scatchard plots for Fe20Tf and for the half-molecules in the presence of excess complementary unlabeled half-molecule corrected for nonspecific binding. The correction is made by taking the slope of the line at the 4 highest concentrations in the binding curve (Fig. 5A) and using it to correct the experimental points for nonsaturable binding. At concentrations between 1 and 60 p~, the nonspecific binding ranges between 2.5 and 51.3% for Fe,OTf, between 2. 8  Extraction of 59Fe into Heme-By following the protocol of Teale (30) as described by Keung and Azari (18), the amount of "Fe incorporated into heme was determined after a 30-min incubation with 10 p~ of the various labeled samples. In the heme containing X-butanone extract, 80.7, 97.8, 96.0, and 88.8% of the total cell-associated 59Fe cpm were found for 5gFez"51-OTf, 59Fe'251-OTf/2N + 59Fe'251-OTf/2C, 5gFe'251- OTf/2N + FeOTf/2C, and FeOTfl2N + 5gFe'z51-OTf/2C, respectively. In other words, of the amount of 59Fe associated with the cells, a similar per cent was extractable into the putative heme fraction.

DISCUSSION
The studies described demonstrate for the first time binding and iron donation in a "reconstituted system which approach those found for diferric transferrin. It is clear from our results that both half-molecules are necessary to effect specific binding to the receptor and the donation of iron. Both uptake and equilibrium binding experiments indicate that the nature of the binding of the half-molecules alone is nonspecific or nonsaturable. This is further confirmed by the apparent inability of the half-molecules alone to donate iron to CERBCs. Whether the combined complementary half-molecules associate prior to binding to the receptor or whether the binding takes place at the receptor site is an important question which we are currently investigating. In either case, the association appears to be very rapid reaching half-maximal within a minute after addition of the unlabeled complementary halfmolecule (see Fig. 4 A , samples 4 and 6). Some interesting differences (Table 11) in binding and iron donation -are observed depending on which half-molecule bears the radiolabel. Although relatively more of the NHzterminal half-molecule than the COOH-terminal half-molecule appears to bind, iron donation by the COOH-terminal half-molecule is almost twice that of the NH,-terminal halfmolecule. The reasons for these differences, particularly in the ability of the half-molecules to donate iron, are not understood. It may be that production of the half-molecule in some way affects the way in which iron is held in the NHzterminal half-molecule which in turn interferes with iron donation. There are a number of indications that the iron in the NHz-terminal half-molecule is more labile than that in the COOH-terminal half-molecule. Thus, we have observed that iron is more readily removed from FeOTf/2N than FeOTf/2C by dialysis against phosphate buffer containing NTA and EDTA. Also, in the EGA medium, the level of iron donation falls in the sample incubated for a longer period of time (Fig. 4B, sample 4 Competition experiments demonstrate that the combined half-molecules are about half as effective as 59Fe21251-OTf in competing with equimolar unlabeled Fe20Tf. This is a significant finding when compared to the results of Keung and Azari (18) in which equimolar FezOTf reduced binding of a half-molecule to 5% of the control. In fact, virtually all of our findings are in conflict with those reported by these authors. In contrasting our results with theirs, we find numerous differences including methods used to prepare the proteins, radiolabeling procedures, experimental procedures, and presentation of results. First, and probably most significant, is their use of subtilisin-derived half-molecules. Furthermore, in presenting their work, they do not show uptake studies in which both binding and iron donation in the same cells were examined. It is interesting to note that significant iron uptake from the half-molecules is found in their experiments at concentrations of protein which are higher than the maximum shown in their binding studies. Iron incorporation from halfmolecules into heme does not level off even a t a concentration of 37.5 KM which is 15 times higher than the highest concentration shown for binding. In view of these results, there is some question as to whether iron from the half-molecules might be transported into the cells via residual endogenous OTf. Our experiments show that with the procedure followed by Keung and Azari to eliminate endogenous OTf there may still be a considerable amount remaining. In our own experiments, we feel that the failure of the half-molecules alone to transport iron serves as a control for the combined halfmolecule results. We are quite confident that the combined half-molecules are not operating by interacting with OTf endogenously present in the system. Furthermore, in our experiments, similar amounts of radiolabeled iron can be extracted into the putative heme fraction from Fe20Tf and the various combined half-molecules.
Therefore, the iron taken up by the cells under these circumstances is specifically available for heme synthesis.
Our first goal in setting up equilibrium binding studies was to try to establish conditions under which the system is at equilibrium. Most past studies have been satisfied with what is described as a "steady state." Recent articles (31, 32) have pointed out that data from such studies are inherently incorrect. Most equilibrium studies, in addition, fail to include a large enough range of concentrations of free ligand to render the resulting plot meaningful (31). Two additional points deserve comment. It is now well established that apotransferrin and diferric transferrin have significantly different binding properties (33-38). Furthermore, there is mounting evidence that, once bound to the receptor, diferric transferrin is internalized by endocytosis, iron is removed in "endosomes," and apotransferrin still associated with the receptor is returned to the surface (39)(40)(41)(42)(43)(44). Although there is not complete agreement that internalization takes place (26,38,45), if it does, the situation may be too complex to be described by Scatchard analysis. Regardless of the details of the process, the use of DNP in our equilibrium binding studies prevents the removal of iron (and probably endocytosis (46)). Under these conditions, a Scatchard plot is appropriate because binding is reversible and equilibrium can be attained. Although DNP reduces apparent binding by 25% and may have other unknown effects, we believe that relative numbers for OTf uerszu the half-molecules are valid. The Scatchard analysis is important in yielding estimates of nonspecific binding which is approximately the same for OTf and the combined half-molecules. If the combined half-molecules had significantly higher nonspecific binding than OTf, there would be concern about the data obtained in the uptake experiments. The apparent binding constants derived from the Scatchard plots are of interest. Again, although the absolute numbers may not be correct, the relative numbers are probably valid. Fig. 5B, in which the amount of ligand bound is plotted against the log of the concentration of free ligand, is included to show that roughly sigmoidal curves, which pass well through an inflection point at approximately half the number of sites found at saturation, are obtained. This is the expectation for a set of n identical binding sites (31). The corrected data, when plotted according to the method of Scatchard (29) for radiolabeled FeOTf/2N in the presence of excess unlabeled FeOTf/BC or labeled FeOTfl2C in the presence of excess unlabeled FeOTflPN, yield straight lines (Fig. 7). The number of binding sites are 92.7 and 87.0% of the number of sites/cell found for Fe20Tf. The lower slopes for the combined halfmolecules reveal somewhat lower binding constants consistent with the competition experiments. Furthermore, the presence of excess unlabeled complementary half-molecules appears to facilitate increased binding of the labeled species. The shape of the curves found in the case of equimolar amounts of the two half-molecules is suggestive of positive binding cooperative (47). It is not until concentrations of about 3 p~ that the curve meets that found for the labeled half-molecule in the presence of excess unlabeled species. The molecular basis for this apparent cooperativity is presently under investigation.
In conclusion, we feel that further studies with the halfmolecules may yield important insights into the interaction of transferrin with its receptor. By differential labeling of the two half-molecules with for example 55Fe and "Fe more information could be obtained as to the flow of iron into the cell. Furthermore, preliminary experiments with apo-halfmolecules have yielded interesting results in regard to both iron uptake and effects on binding. The interacting complementary half-molecule provide a unique opportunity to dissect the effects of each iron-binding domain on the other.

Binding and
Iron Donation by Ovotransferrin Half-molecules