The Position of Arginine 124 Controls the Rate of Iron Release from the N-lobe of Human Serum Transferrin A STRUCTURAL STUDY*

Human serum transferrin (hTF) is a bilobal iron-bind-ing and transport protein that carries iron in the blood stream for delivery to cells by a pH-dependent mechanism. Two iron atoms are held tightly in two deep clefts by coordination to four amino acid residues in each cleft (two tyrosines, a histidine, and an aspartic acid) and two oxygen atoms from the “synergistic” carbonate anion. Other residues in the binding pocket, not directly coordi-nated to iron, also play critical roles in iron uptake and release through hydrogen bonding to the liganding residues. The original crystal structures of the iron-loaded N-lobe of hTF (pH 5.75 and 6.2) revealed that the synergistic carbonate is stabilized by interaction with Arg-124 and that both the arginine and the carbonate adopt two conformations (MacGillivray, R. T. A., Moore, S. A., Chen, J., Anderson, B. F., Baker, H., Luo, Y. G., Bewley, M., Smith, C. A., Murphy, M. E., Y., Mason, A. B., Woodworth, R. C., E.

Human serum transferrin (hTF) 1 is a member of the transferrin family of iron-binding proteins, which includes ovotrans-ferrin (oTF), found in avian egg white, lactoferrin (LTF), found in milk, tears, and other bodily secretions, and melanotransferrin, found on the surface of melanocytes (1,2). The hTF binds iron reversibly in blood plasma and transports it to cells requiring iron. Full-length transferrin molecules (ϳ80 kDa) consist of a single polypeptide chain folded into two similar lobes, the N-lobe and C-lobe. The two lobes display significant sequence similarity and appear to have evolved from the duplication of an ancestral gene coding for a protein with a single metal-binding site (3). Each homologous lobe contains an ironbinding site deep within a cleft that subdivides it into two dissimilar domains (designated NI and NII domains for the N-lobe and CI and CII for the C-lobe). Iron is bound in a distorted octahedral coordination involving four amino acid ligands and two oxygen atoms from a synergistically bound carbonate ion. The synergistic relationship of metal and anion refers to the fact that neither binds tightly in the absence of the other (4). This synergy is a unique characteristic of the transferrin family and is not observed in other metal-binding proteins. Although hTF, oTF, and LTF share identical iron-binding ligands and display high sequence homology, substantial differences in the binding affinity for iron both within and between the TFs are found and are not well understood (5)(6)(7)(8)(9)). An obvious experimental approach to explore these differences involves the mutation of specific residues followed by assessment of the changes in function.
A critical function of hTF in serum is to deliver iron to actively dividing cells. Diferric hTF binds with high affinity to specific receptors on the surface of cells (10). The apo-or irondepleted hTF is poorly recognized by the receptor at physiological pH. The hTF-receptor complex is endocytosed; within the endosome, a proton pump results in a drop in the pH to ϳ5.6, triggering iron release. This step is not well defined despite intensive research efforts. The entire cycle occurs in 2-3 min (11).
Iron binding and release is characterized by a large conformational change. When iron is released, the two domains move away from each other by means of a hinge located in two antiparallel ␤-strands that lie behind the iron-binding site in each lobe. In the N-lobe of transferrin, one domain rotates 63°r elative to the other about this hinge (12,13). The precise mechanism for this movement remains unclear, but protonation of several key amino acid residues appears to be critical to the release of iron (see below).
Many studies have clearly demonstrated that an arginine located near the iron-binding site (Arg-124 in hTF/2N) stabilizes the synergistic carbonate anion. This arginine is highly conserved in each lobe of all mammalian TFs. As detailed previously (4), those lobes with amino acids other than arginine are unable to bind iron with high affinity and specificity. Mutational studies in both hTF and LTF further confirm the important role of the arginine in iron release (14 -17); all mutants feature accelerated rates of iron release. The two high resolution structures (1.6 and 1.8 Å) of the Fe(III) form of hTF/2N revealed the presence of electron density indicative of positional disorder near the metal-binding sites of both crystal forms. The density was fitted by placement of the synergistic carbonate and the side chain of Arg-124 into two positions. In the A or "near" conformer, the carbonate is fully engaged in binding to the iron and is stabilized by bonds to the NE and NH 2 atoms of Arg-124, the OG1 atom of Thr-120, and the main chain amide nitrogen atoms of Ala-126 and Gly-127, which reside on helix 5 (Fig. 1A). This form is found in both lobes of oTF and LTF. In contrast, in the B or "far" conformer, the carbonate has rotated 30°, Arg-124 has moved away from the site, and the carbonate is detached from helix 5 (Fig. 1B). We hypothesized that protonation of the carbonate (resulting in bicarbonate) might be responsible for these changes in the position of the arginine and that this protona-tion could well be the first step in iron release, an idea originally suggested by Aisen and Leibman in 1973 (18). Obviously, changes in access to carbonate could have a profound effect on the release rate. In the wild-type N-lobe of hTF, the carbonate is surrounded by a network of ordered water molecules that may function in the transport of protons from the outside of the protein into the iron-binding site. Of interest is that no heterogeneity has been observed in the position of the arginine and carbonate in either lobe of any of the other TF structures reported to date. These include many different LTF structures (19,20), oTF (21), rabbit and pig serum transferrin (22), and the iron-containing C-lobe of hTF (23).
The current study was undertaken to determine the significance of the alternate positions of Arg-124 in the function of the N-lobe of hTF. It is important to mention that the present study includes the structure of wild-type hTF/2N crystallized at pH 7.7 as a control and that it also features alternate positions for both the carbonate and Arg-124. This implies that the original observation was not directly due to the lower pH used in these earlier studies as postulated previously (12). Tyrosine 45 resides at the lip of the iron-binding cleft and is hydrogen-bonded to a water molecule; it was mutated to a glutamic acid residue (Y45E). We hypothesized that the presence of the negatively charged glutamic acid might force Arg-124 into a single position, namely the B or far conformer, allowing easier access to the site. By replacing the tyrosine with glutamic acid in the wild-type structure, the carboxyl head group is the proper distance from Arg-124 to produce a salt bridge ( Fig. 2A).
Alternatively Leu-66, which in the closed iron-containing N-lobe resides near Arg-124 on the edge of the solvent cavity adjacent to the iron-binding site, was converted to a tryptophan (L66W) with the premise that substitution of the bulky, hydrophobic Trp residue would force Arg-124 into the A or near conformer and thus restrict access to the synergistic anion (Fig.  2B). Functional and structural data appear to support these hypotheses and confirm the importance of the alternate conformations of carbonate and Arg-124 in the mechanism of iron release from the N-lobe of hTF.

Materials
All chemicals used were of reagent quality. Stock solutions of HEPES, MES, and other buffers were prepared by dissolving the anhydrous salts in Milli-Q (Millipore) purified water, and adjusting the pH to the desired values with 1 N NaOH or HCl. EDTA was purchased from Mann Research Laboratories, Inc.; nitrilotriacetate was from Sigma, and Tiron was from Fisher. Tiron stock solutions were prepared by dissolving the Tiron in the appropriate buffers and adjusting the pH to the desired values with 1 N NaOH. Centricon 10 microconcentrators were from Amicon. Polyethylene glycol 3350 was from Hampton Research, Inc.

Molecular Biology
The mutant Y45E was introduced into the N-lobe of transferrin using a polymerase chain reaction-based mutagenesis procedure (24). The following synthetic oligonucleotide was used to introduce the mutations (the mutagenic nucleotides are in bold type and underlined): Y45E: 5Ј-AAA GCC TCC GAA CTT GAT TGC-3Ј.
The conditions for the PCR reactions were as follows: denaturation at 94°for 15 s, annealing at 50°for 30 s, and extension at 72°for 30 s.
Step I of the PCR mutagenesis procedure consisted of 30 cycles, step II consisted of 1 cycle, and step III consisted of 30 cycles (24). The nucleotide sequence of the insert confirmed the introduction of the specific mutation. The mutated hTF/2N cDNA was excised from the Bluescript vector, the ends were made blunt, and the fragment was ligated into the SmaI site of the pNUT expression vector. Restriction endonuclease mapping was used to confirm the correct orientation.
The L66W mutation was introduced into the transferrin N-lobe using the QuikChange mutagenesis kit (Stratagene). The mutation was made directly in the transferrin N-lobe cDNA sequence cloned in the pNUT vector. Two complimentary mutagenic oligonucleotide primers contain- All structural figures were drawn using Bobscript (37,38) and Raster3D (39,40).
The conditions for the PCR reaction were as follows: 95°for 30 s followed by 18 cycles with denaturation at 95°for 30 s, annealing at 55°f or 1 min, and extension at 68°for 13 min. To determine the presence of the correct mutation and absence of other mutations, the complete sequence of the transferrin cDNA and flanking pNUT sequence was determined prior to the introduction of the plasmid into baby hamster kidney cells.

Recombinant Protein Production and Purification
The production and purification of hTF/2N and mutants of hTF/2N in baby hamster kidney cells using the pNUT expression vector system has been described previously in detail (25,26). Briefly, the recombinant hTF/2N that is secreted into the tissue culture medium is saturated with iron and exchanged into 5 mM Tris-HCl buffer, pH 8.0, using a spiral cartridge concentrator. A Poros 50 HQ anion-exchange column is used to eliminate most of the serum albumin and all of the phenol red from the sample. Pooled fractions are concentrated and applied to a Sephacryl S-200 HR column (5 ϫ 80 cm) equilibrated and run in 0.1 M ammonium bicarbonate. Following passage through a 0.2-m syringe filter and concentration, the samples are stored at Ϫ20°C in 0.1 M ammonium bicarbonate. Purity of the recombinant protein was determined using SDS-polyacrylamide gel electrophoresis.

Kinetics of Iron Removal
Iron removal from the wild-type N-lobe and the mutants (ϳ40 M) was measured using the chelator Tiron (12 mM) in 50 mM HEPES at pH 7.4 and 25°C. The reaction was monitored by following an increase in absorbance at 480 nm for the iron-Tiron complex formation. For experiments at pH 5.6, the chelator EDTA was used to remove iron at a concentration of 4 mM in 50 mM MES. In this case, the reaction was monitored by the decrease in absorbance at 470 or 293 nm, which follows the release of iron from the protein. For slower release rates, a Cary 100 spectrophotometer (Varian) was used. For faster rates, iron release kinetics were measured using an OLIS RSM-1000 stop-flow spectrophotometer (On-Line Instrumentation Systems, Inc). One syringe contained the protein sample in water, and the other contained 8 mM EDTA in 100 mM MES buffer, pH 5.6. Absorbance spectra were collected 5 ms after mixing and continued for at least four half-lives.

pH Dependence on Iron Release
The retention of iron as a function of pH was measured for each mutant. Aliquots of iron-saturated protein (ϳ50 M) in 100 mM ammonium bicarbonate were incubated in a buffer containing 33.3 mM HEPES, MES, and sodium acetate adjusted to the appropriate pH (between 3 and 8) with either 1 N NaOH or 9 lacial acetic acid and maintained at 4°C for a period of 1 week to allow each sample to reach equilibrium. The percentage of iron remaining bound to the transferrin samples was determined by measuring the absorbance at the visible absorption maxima and comparing this absorbance to the fully ironloaded protein. The pH was measured on identical aliquots at the end of 1 week. The data were plotted and analyzed using Origin software (Microcal).

Crystallization of hTF/2N and the Two Mutants
Recombinant hTF/2N and both mutants were crystallized using the sitting drop method at 20°C. Protein (35 mg/ml) in 0.1 M ammonium bicarbonate was mixed with an equal amount of the reservoir solution composed of 0.2 M potassium acetate (pH 7.7) and 20% polyethylene glycol 3350. In all cases, large red crystals (2.0 mm ϫ 0.9 mm ϫ 0.4 mm) formed in 5-14 days. All proteins were essentially isomorphous with wild-type hTF/2N (Protein Data Bank accession number 1A8E), showing similar cell dimensions and crystallizing in the orthorhombic space group P2 1 2 1 2 1 with one molecule in each asymmetric unit (Table I).

Data Collection, Structure Determination, and Refinement
Data for hTF/2N and the two mutants were collected at room temperature using a Mar345 image plate detector on a Rigaku RU-300 generator with a copper rotating anode. All data sets were indexed using DENZO (27) and scaled and merged using SCALEPACK (27). Statistics are given in Table I. For all structures, the wild-type pH 5.75 structure (Protein Data Bank accession number 1A8E) was used with molecular replacement. In all cases Arg-124, carbonate, iron, and waters were removed from the search model to eliminate any bias from these atoms Wild-type hTF/2N, pH 7.7-Since the unit cell was identical to the previously solved wild-type hTF/2N, the rigid-body alignment routine of CNS (in which the NI and NII domains were also allowed to move independently) was used to solve the wild-type hTF/2N, pH 7.7, structure (28). This reduced the R-factor to 0.30 for data from 30 to 2.05 Å. The model was refined using successive rounds of simulated annealing, occupancy, and B-factor refinement. This was followed by revision of the model based upon map interpretation using an SGI work station with the graphics program O (29,30). In the later stages of refinement, Arg-124, carbonate, and solvent molecules were added and allowed to move freely. Conformations of Arg-124 and carbonate were determined by fitting observed F o Ϫ F c electron density maps. Only residues 3-331 from the polypeptide chain were observed in the electron density and deposited in the final model (Protein Data Bank accession number 1N84).
Y45E-The structure of the Y45E mutant was solved by using the molecular replacement routines in CNS. The cross-rotation search yielded a single peak 8.8 above the mean, and the translational search found a peak 3.8 above the mean. Rigid-body refinement reduced the R-factor to 0.31 for data from 30 to 2.1 Å. Refinement was carried out as described above; again, only residues 3-331 were deposited in the final model (Protein Data Bank accession number 1N7X).
L66W-As described for the wild-type pH 7.7 structure above, the rigid-body alignment routine of CNS was used to solve the L66W mutant; this reduced the R-factor to 0.29 for data from 30 to 2.1 Å. Refinement was carried out as described above, again with only residues 3-331 deposited in the final model (Protein Data Bank accession number 1N7W).

RESULTS AND DISCUSSION
Iron Binding and Release Kinetics-Iron-loaded hTF/2N has characteristic absorption minima ( min ) and maxima ( max ) in the visible range that give it its distinct red color. These values, together with the ratios of A max /A min and A 280 /A max , reflect the metal binding properties of the protein. As indicated in Table  II, the absorption maximum of both mutant transferrins shows a small blue shift as compared with wild-type hTF/2N. The other spectral properties are essentially unchanged from those of the wild-type protein. This suggests that the position of the iron-binding ligands of these mutants is not greatly perturbed by the introduction of these mutations. The spectral properties of the R124A mutant are included for the purpose of comparison; the R124A mutant features a significant downshift in the absorbance maxima, indicating a disturbance in the relationship of the iron with the two liganding tyrosine residues, which are mainly responsible for the absorbance maximum (31).
Rate constants for the release of iron from the wild-type hTF/2N and the two mutants were determined at both pH 7.4 and 5.6 in the presence and absence of salt. As shown in Table  III, the release rate of iron from the Y45E mutant is 30-fold faster than that found for the wild-type protein at both pH values. At pH 7.4, chloride slows iron release, a result that is also observed with wild-type hTF/2N. At pH 5.6, however, release from the Y45E mutant is accelerated only 2-fold in the presence of chloride ions versus the 4 -5-fold acceleration observed for wild-type hTF/2N. At pH 7.4, no iron was released from the L66W mutant under the identical conditions used to measure iron release from wild-type hTF/2N and the Y45E mutant. At pH 5.6, the rate of release was 21 times slower than that measured for wild-type hTF/2N. Again, the rates of release for the R124A mutant are included for comparison and show an even faster rate of release at both pH 7.4 (100-fold) and 5.6 (50-fold) relative to the wildtype hTF/2N. At both pH values, the salt effect on the R124A mutant is reduced as compared with wild-type hTF/2N.
In the hTF N-lobe, the presence of anions (chloride in this case) exerts a suppressing effect on iron release due to competition between chloride and the chelator for a site or sites near the metal center (16). In addition, we note that as the pH is  lowered, the strength of chloride binding increases in a linear fashion (16). Lowering of the pH is also associated with the protonation of various residues, leading to opening of the cleft and exposure of additional side chains to which chloride can attach, thereby further promoting and maintaining cleft opening. We have found that any mutation that affects any of the anion-binding residues leads to a reduction in the effect of chloride on the rate. Our studies have identified Arg-124, Lys-206, and Lys-296 as major anion-binding residues (32). To date, no single mutation has resulted in a complete abolition of the chloride effect, implying that multiple residues are involved. In the current study, both mutants impact Arg-124, which probably accounts for the muted chloride effect. The data presented here are consistent with previous work in which Arg-124 was mutated to serine, lysine, alanine, or glutamic acid (4,14,15,17). All of these studies verified the critical role of the arginine in anchoring the synergistic carbonate anion. In a recent study by Zak et al. (33), the threonine residue at position 120, which also helps to stabilize the synergistic carbonate anion, was mutated to an alanine. The T120A hTF/2N mutant featured weakened iron binding as indicated by iron release studies, although the effect was not as great as that observed for the R124A mutant.
The L66W mutant features two rates for iron release in the presence of 50 mM chloride at pH 5.6. One rate is three times faster than that found in the absence of salt, whereas the other is six times slower. Biphasic iron release has been observed for the N-lobe of oTF at pH 5.6, and a model has been offered and mathematically tested (9). Muralidhara and Hirose (9) propose that the two rates reflect release from two different protein conformations, namely a domain-closed conformer and an anion-induced different conformer. The rates are very dependent on the nature of the anion; there is a competition between binding directly to the iron ligands versus binding to a specific anion site or sites.
pH Titration of Iron Removal-The effect of pH on iron release can be determined by an equilibrium experiment; the results of such an experiment are presented in Fig. 3. It is important to note that no chelator is present in these samples. There is a slow protonation of the iron-binding residues, leading to iron release at acidic pH even in the absence of chelators (34). As estimated from the pH profile, approximately half of the iron is lost at a pH of 4.77 Ϯ 0.21 (n ϭ 4) for wild-type hTF/2N. The Y45E mutant loses 50% of its iron at pH 3.99 Ϯ 0.03 (n ϭ 2); the L66W mutant loses half of its iron at pH 4.91 Ϯ 0.08 (n ϭ 2). These results seem at odds with the kinetic results; the kinetically more stable L66W mutant has a less stable thermodynamic profile and loses iron in a very narrow pH range. In contrast, the less kinetically stable WT and the Y45E mutant show a more gradual loss of iron over a wider pH range. The kinetically least stable Y45E mutant in fact is the most resistant to iron loss with a pH almost a full pH unit below the L66W mutant (see below).
Structures of hTF/2N, Y45E, and L66W-The N-lobe of hTF and both mutants were crystallized under identical conditions. The final values R and R free were 19.8 and 22.9% for hTF/2N, 19.6 and 23.8% for the L66W mutant, and 19.3 and 24.2% for the Y45E mutant (Table I).
To determine whether the two conformations of Arg-124 and carbonate seen in the previous structures were a function of the low crystallization pH values (pH 5.75 and 6.1), wild-type hTF/2N was crystallized at pH 7.7. The wild-type hTF/2N structure at pH 7.7 was then refined to 2.05 Å and was directly compared with the wild-type structure solved at pH 5.75 (Protein Data Bank accession number 1A8E). Superposition of this wild-type pH 7.7 hTF/2N structure with the pH 5.75 wild-type structure results in an overall root mean squared deviation of only 0.45 Å, clearly demonstrating that no large domain shifts occur as a result of the change in pH. One difference in the pH 7.7 structure is that residues Ser-44, Ser-248, Thr-250, and Ser-255 refine to a single conformation as compared with two conformations observed at pH 5.75. Significantly, both Arg-124 and the synergistic carbonate display two conformations as observed in the original structural data (Table IV). At this higher pH, the occupancy of each conformer is approximately equal for Arg-124, whereas the carbonate refines to 0.65 and 0.35 for the A and B conformer, respectively. In addition, although the A conformation of Arg-124 is identical to the A conformer (root mean squared difference ϭ 0.09 Å) of the pH 5.75 structure, there is a change in the position of the B conformer (root mean squared difference ϭ 0.96 Å). In the pH 7.7 structure, the NE imino group of Arg-124 is rotated nearly 180°t oward the iron-binding site with the NH1 and NH 2 amino groups facing away from the carbonate (Fig. 1B). This position is almost identical to that observed for the B conformer in the   FIG. 3. pH titration of WT hTF/2N and two mutants. Iron release for each mutant was measured over a pH range (pH 3-8). The percentage of iron remaining bound to each protein was determined by measuring the absorbance at the visible absorption maxima (472 nm for WT hTF/2N, 468 nm for the Y45E mutant, and 470 nm for the L66W mutant) as a percentage of the fully iron-loaded protein. The Y45E mutant (circles) showed iron release at a lower pH than WT hTF/2N (squares), whereas the L66W mutant (triangles) released at a higher pH as compared with WT hTF/2N. Critical Role of the Anion-binding Site pH 6.1 structure (Protein Data Bank accession number 1A8F) (12).
In the Y45E structure, the side chain of Arg-124 is found strictly in the B or far conformation. In contrast, the carbonate ion bound to the iron clearly shows two conformations (Fig. 4). The glutamic acid residue at position 45 in the NI domain forms a salt bridge with Arg-124 located in the NII domain, locking the arginine into this single position. Arg-124 is swung away such that only the NE amino group can form a hydrogen bond with the O3 oxygen atom of the carbonate anion. This is in contrast to the R124A mutant in which all hydrogen bonding is abolished by the mutation. We believe that this could explain why the rate of iron release for the R124A mutant is three times greater than that found for the Y45E mutant at pH 7.4. In the Y45E mutant, additional positive density was observed near the iron-binding site. The density could not be fit to a second arginine conformation due to distance constraints. Fitting with hydrogen bond distance and angle constraints, this additional density was modeled as a water molecule (509), forming a hydrogen bond with the O1 and O3 carbonate oxygens. We believe that this water molecule is part of the network of ordered water molecules that function to transport protons into the binding site. With the arginine in the far position and this extra water molecule in its place, protons have access to the carbonate ion destabilizing its interaction with iron, and the chelator has better access to the iron. This finding provides an explanation for the kinetic data; access to the carbonate anion is unobstructed, leading to an acceleration of the rate of release of the iron (Table III). In the absence of chelator, as in the pH studies in Fig. 3, the salt bridge between Arg-124 and Glu-45 (which reside in opposing domains) stabilizes the closed structure relative to the wild type (see below).
In contrast to the Y45E mutant, replacement of the leucine at position 66 with a tryptophan to make the L66W mutant results in the complete shutdown of iron release at physiological pH. Structurally, leucine 66 sits on the periphery of a solvent cavity adjacent to the iron-binding site. In the wild-type molecule, neither Arg-124 conformer interacts directly with leucine 66. Upon mutation to a tryptophan, however, the large indole ring of Trp-66 projects into this solvent cavity. The crystal structure shows that this larger side chain stacks with the side chain of Tyr-45 and occupies the space where the B conformer of Arg-124 normally resides in the wild-type structure (Fig. 5). By restricting access, the side chain can only adopt the conformation of the wild-type A or near conformer. Significantly, the synergistic carbonate anion also displays a single conformation (Table IV). The NH 2 and NE atoms of Arg-124 form hydrogen bonds to the O3 atom of the bound carbonate, as found in all other TF structures. We speculate that by trapping the arginine close to the iron, the synergistic anion is shielded from protonation by the solvent. Chicken oTF contains a glutamine residue at the position equivalent to 66 and also displays a single conformation of arginine and carbonate (21). It is tempting to speculate that protection may contribute to the slower iron release of oTF/2N as compared with WT hTF /2N (9, 35). In human LTF, two phenylalanine residues at positions 63 and 183 occupy the space that would be occupied  (Table IV). by the B conformer of Arg-124. Likewise, the restricted movement of the equivalent LTF Arg (Arg-121) may contribute to its greater kinetic stability. Of great interest is the fact that both rat and mouse serum TFs actually contain a tryptophan residue at position 66. Work by Morgan (36) confirms the prediction that a significantly reduced iron release rate is found for both oTF and rat TF as compared with hTF under identical conditions. The two mutations in the current study reside in residues that are located in the NI domain, whereas the Arg-124 resides in the NII domain. One might predict that this inter-domain interaction, particularly in the case where the glutamic acid forms a salt bridge with the arginine, should stabilize the closed form of the protein. This appears to be the case in the equilibrium studies in which iron loss as a function of pH was measured. With no chelator present, it takes a significantly lower pH to release 50% of the iron from the Y45E mutant (Fig.  3). As mentioned above, these results differ from the results in the kinetic studies. Clearly in the Y45E mutant, increased access to the carbonate (when a chelator is present) is a stronger factor than the interaction between domains in dictating the rate of iron release. This is in contrast to the two lysines (Lys-206 and Lys-296) that also reside in the opposing domains and that are critically important in iron release (32). In the iron form of hTF/2N, one member of the pair is positively charged, whereas the other is uncharged, allowing formation of a shared hydrogen bond between them. As the pH is lowered (as in the endosome), protonation results in electrostatic repulsion of these two lysines, providing a force to open the cleft, thereby exposing the bound iron and allowing the release of iron. Mutation of either lysine residue to alanine greatly retards the release of iron, demonstrating that the lysines act as a trigger rather than as a lock. In addition, mutation of either of the positively charged lysine residues to a negatively charged glutamic acid appears to lock the N-lobe in the closed conformation, making release very slow (32).
In conclusion, when combined with previous data in which the arginine at position 124 was mutated, the information from the present study confirms the importance of Arg-124 in stabilization of the synergistic anion. The present work allows us to suggest that movement of this conserved arginine is the critical first step in the release of iron from the N-lobe of hTF. The fact that no other TF structures to date show alternative positions for the arginine and the carbonate raises the question of whether the N-lobe of hTF is unique in this regard. It is possible that the absence of the C-lobe allows the movement to take place and to be more easily captured during crystallization. Current efforts are focused on obtaining crystals of recombinant diferric hTF to test this idea.