Phylogenetic and sequence analyses of insect transferrins suggest that only transferrin 1 has a role in iron homeostasis

Iron is essential to life,but surprisingly little is known about how iron is managed in nonvertebrate animals.In mammals,the well-characterized transferrins bind iron and are involved in iron transport or immunity,whereas other members of the transferrin family do not have a role in iron homeostasis.In insects,the functions of transferrins are still poorly understood.The goals of this project were to identify the transferrin genes in a diverse set of insect species,resolve the evolutionary relationships among these genes,and predict which of the transferrins are likely to have a role in iron homeostasis.Our phylogenetic analysis of transferrins from 16 orders of insects and two orders of noninsect hexapods demonstrated that there are four orthologous groups of insect transferrins.Our analysis suggests that transferrin 2 arose prior to the origin of insects,and transferrins/,i,and 4 arose early in insect evolution.Primary sequence analysis of each of the insect transferrins was used to predict signal peptides,carboxyl-terminal transmembrane regions,GPI-anchors,and iron binding.Based on this analysis,we suggest that transferrins 2,and 4 are unlikely to play a major role in iron homeostasis.In contrast,the transferrin 1 orthologs are predicted to be secreted,soluble,iron-binding proteins.We conclude that transferrin 1 orthologs are the most likely to play an important role in iron homeostasis.Interestingly,it appears that the louse,aphid,and thrips lineages have lost the transferrin 1 gene and,thus,have evolved to manage iron without transferrins.     

Structural insight into the novel iron‐coordination and domain interactions of transferrin‐1 from a model insect,Manduca sexta

Transferrins function in iron sequestration and iron transport by binding iron tightly and reversibly. Vertebrate transferrins coordinate iron through interactions with two tyrosines, an aspartate, a histidine, and a carbonate anion, and conformational changes that occur upon iron binding and release have been described. Much less is known about the structure and functions of insect transferrin‐1 (Tsf1), which is present in hemolymph and influences iron homeostasis mostly by unknown mechanisms. Amino acid sequence and biochemical analyses have suggested that iron coordination by Tsf1 differs from that of the vertebrate transferrins. Here we report the first crystal structure (2.05 Å resolution) of an insect transferrin. Manduca sexta (MsTsf1) in the holo form exhibits a bilobal fold similar to that of vertebrate transferrins, but its carboxyl‐lobe adopts a novel orientation and contacts with the amino‐lobe. The structure revealed coordination of a single Fe³⁺ ion in the amino‐lobe through Tyr90, Tyr204, and two carbonate anions. One carbonate anion is buried near the ferric ion and is coordinated by four residues, whereas the other carbonate anion is solvent exposed and coordinated by Asn121. Notably, these residues are highly conserved in Tsf1 orthologs. Docking analysis suggested that the solvent exposed carbonate position is capable of binding alternative anions. These findings provide a structural basis for understanding Tsf1 function in iron sequestration and transport in insects as well as insight into the similarities and differences in iron homeostasis between insects and humans.     

Delineating the regions of human transferrin involved in interactions with transferrin binding protein B from Neisseria meningitidis

Pathogenic bacteria in the Neisseriaceae possess a surface receptor mediating iron acquisition from human transferrin (hTf) that consists of a transmembrane iron transporter (TbpA) and a surface‐exposed lipoprotein (TbpB). In this study, we used hydrogen/deuterium exchange coupled to mass spectrometry (H/DX‐MS) to elucidate the effects on hTf by interaction with TbpB or derivatives of TbpB. An overall conserved interaction was observed between hTf and full‐length or N‐lobe TbpB from Neisseria meningitidis strains B16B6 or M982 that represent two distinct subtypes of TbpB. Changes were observed exclusively in the C‐lobe of hTf and were caused by the interaction with the N‐lobe of TbpB. Regions localized to the ‘lip’ of the C1 and C2 domains that flank the interdomain cleft represent sites of direct contact with TbpB whereas the peptides within the interdomain cleft that encompass iron binding ligands are inaccessible in the closed (holo) conformation. Although substantial domain separation upon binding TbpB cannot be excluded by the H/DX‐MS data, the preferred model of interaction involves binding hTf C‐lobe in the closed conformation. Alternate explanations are provided for the substantial protection from deuteration of the peptides encompassing iron binding ligands within the interdomain cleft but cannot be differentiated by the H/DX‐MS data.     

Transferrin binding and iron uptake in mouse hepatocytes

Methods were developed for obtaining highly viable mouse hepatocytes in single cell suspension and for maintaining the hepatocytes in adherent static culture. The characteristics of transferrin binding and iron uptake into these hepatocytes was investigated. (1) After attachment to culture dishes for 18–24 h hepatocytes displayed an accelerating rate of iron uptake with time. Immediately after isolation mouse hepatocytes in suspension exhibited a linear iron uptake rate of 1.14·10⁵molecules/cell per min in 5 μM transferrin. Iron uptake also increased with increasing transferrin concentration both in suspension and adherent culture. Pinocytosis measured in isolated hepatocytes could account only for 10–20% of the total iron uptake. Iron uptake was completely inhibited at 4°C. (2) A transferrin binding component which saturated at 0.5 μM diferric transferrin was detected. The number of specific, saturable diferric transferrin binding sites on mouse hepatocytes was 4.4·10⁴±1.9·10⁴ for cells in suspension and 6.6·10⁴±2.3·10⁴ for adherent cultured cells. The apparent association constants were 1.23·10⁷ 1·mol^?1 and 3.4·10⁶ 1·mol^?1 for suspension and cultured cells respectively. (3) Mouse hepatocytes also displayed a large component of non-saturable transferrin binding sites. This binding increased linearly with transferrin concentration and appeared to contribute to iron uptake in mouse hepatocytes. Assuming that only saturable transferrin binding sites donate iron, the rate of iron uptake is about 2.5 molecules iron/receptor per min at 5 μM transferrin in both suspension and adherent cells and increases to 4 molecules iron/receptor per min at 10 μM transferrin in adherent cultured cells. These rates are considerably greater than the 0.5 molcules/receptor per min observed at 0.5 μM transferrin, the concentration at which the specific transferrin binding sites are fully occupied. The data suggest that either the non-saturable binding component donates some iron or that this component stimulates the saturable component to increase the rate of iron uptake. (4) During incubations at 4°C the majority of the transferrin bound to both saturable and nonsaturable binding sites lost one or more iron atoms. Incubations including 2 mM α,α′-dipyridyl (an Fe¹¹ chelator) decreased the cell associated ⁵⁹Fe at both 4 and 37°C while completely inhibiting iron uptake within 2–3 min of exposure at 37°C. These observations suggest that most if not all iron is loosened from transferrin upon interaction of transferrin with the hepatocyte membrane. There is also greater sensitivity of ⁵⁹Fe uptake compared to transferrin binding to pronase digestion, suggesting that an iron acceptor moiety on the cell surface is available to proteolysis.     

The genes that encode the gonococcal transferrin binding proteins,TbpB and TbpA,are differentially regulated by MisR under iron‐replete and iron‐depleted conditions

Neisseria gonorrhoeae produces two transferrin binding proteins, TbpA and TbpB, which together enable efficient iron transport from human transferrin. We demonstrate that expression of the tbp genes is controlled by MisR, a response regulator in the two‐component regulatory system that also includes the sensor kinase MisS. The tbp genes were up‐regulated in the misR mutant under iron‐replete conditions but were conversely down‐regulated in the misR mutant under iron‐depleted conditions. The misR mutant was capable of transferrin‐iron uptake at only 50% of wild‐type levels, consistent with decreased tbp expression. We demonstrate that phosphorylated MisR specifically binds to the tbpBA promoter and that MisR interacts with five regions upstream of the tbpB start codon. These analyses confirm that MisR directly regulates tbpBA expression. The MisR binding sites in the gonococcus are only partially conserved in Neisseria meningitidis, which may explain why tbpBA was not MisR‐regulated in previous studies using this related pathogen. This is the first report of a trans‐acting protein factor other than Fur that can directly contribute to gonococcal tbpBA regulation.     

Amonabactin-mediated iron acquisition from transferrin and lactoferrin by <Emphasis Type="Italic">Aeromonas hydrophila</Emphasis> : direct measurement of individual microscopic rate constants

 The effectiveness and mechanism of iron acquisition from transferrin or lactoferrin by Aeromonas hydrophila has been analyzed with regard to the pathogenesis of this microbe. The ability of A. hydrophila's siderophore, amonabactin, to remove iron from transferrin was evaluated with in vitro competition experiments. The kinetics of iron removal from the three molecular forms of ferric transferrin (diferric, N- and C-terminal monoferric) were investigated by separating each form by urea gel electrophoresis. The first direct determination of individual microscopic rates of iron removal from diferric transferrin is a result. A. hydrophila 495A2 was cultured in an iron-starved defined medium and the growth monitored. Addition of transferrin or lactoferrin promoted bacterial growth. Growth promotion was independent of the level of transferrin or lactoferrin iron saturation (between 30 and 100%), even when the protein was sequestered inside dialysis tubing. Siderophore production was also increased when transferrin or lactoferrin was enclosed in a dialysis tube. Cell yield and growth rate were identical in experiments where transferrin was present inside or outside the dialysis tube, indicating that binding of transferrin was not essential and that the siderophore plays a major role in iron uptake from transferrin. The rate of iron removal from diferric transferrin shows a hyperbolic dependence on amonabactin concentration. Surprisingly, amonabactin cannot remove iron from the more weakly binding N-terminal site of monoferric transferrin, while it is able to remove iron from the more strongly binding C-terminal site of monoferric transferrin. Iron from both sites is removed from diferric transferrin and it is the N-terminal site (which does not release iron in the monoferric protein) that releases iron more rapidly! It is apparent that there is a significant interaction of the two lobes of the protein with regard to the chelator access. Taken together, these results support an amonabactin-dependent mechanism for iron removal by A. hydrophila from transferrin and lactoferrin. The implications of these findings for an amonabactin-dependent mechanism for iron removal by A. hydrophila from transferrin and lactoferrin are discussed. Received: 8 August 1999 / Accepted: 22 October 1999     

The structure and evolution of the murine inhibitor of carbonic anhydrase: A member of the transferrin superfamily

The original signature of the transferrin (TF) family of proteins was the ability to bind ferric iron with high affinity in the cleft of each of two homologous lobes. However, in recent years, new family members that do not bind iron have been discovered. One new member is the inhibitor of carbonic anhydrase (ICA), which as its name indicates, binds to and strongly inhibits certain isoforms of carbonic anhydrase. Recently, mouse ICA has been expressed as a recombinant protein in a mammalian cell system. Here, we describe the 2.4 Å structure of mouse ICA from a pseudomerohedral twinned crystal. As predicted, the structure is bilobal, comprised of two α‐β domains per lobe typical of the other family members. As with all but insect TFs, the structure includes the unusual reverse γ‐turn in each lobe. The structure is consistent with the fact that introduction of two mutations in the N‐lobe of murine ICA (mICA) (W124R and S188Y) allowed it to bind iron with high affinity. Unexpectedly, both lobes of the mICA were found in the closed conformation usually associated with presence of iron in the cleft, and making the structure most similar to diferric pig TF. Two new ICA family members (guinea pig and horse) were identified from genomic sequences and used in evolutionary comparisons. Additionally, a comparison of selection pressure (dN/dS) on functional residues reveals some interesting insights into the evolution of the TF family including that the N‐lobe of lactoferrin may be in the process of eliminating its iron binding function.     

An iron-dependent and transferrin-mediated cellular uptake pathway for plutonium

Plutonium is a toxic synthetic element with no natural biological function, but it is strongly retained by humans when ingested. Using small-angle X-ray scattering, receptor binding assays and synchrotron X-ray fluorescence microscopy, we find that rat adrenal gland (PC12) cells can acquire plutonium in vitro through the major iron acquisition pathway--receptor-mediated endocytosis of the iron transport protein serum transferrin; however, only one form of the plutonium-transferrin complex is active. Low-resolution solution models of plutonium-loaded transferrins derived from small-angle scattering show that only transferrin with plutonium bound in the protein's C-terminal lobe (C-lobe) and iron bound in the N-terminal lobe (N-lobe) (Pu(C)Fe(N)Tf) adopts the proper conformation for recognition by the transferrin receptor protein. Although the metal-binding site in each lobe contains the same donors in the same configuration and both lobes are similar, the differences between transferrin's two lobes act to restrict, but not eliminate, cellular Pu uptake.     

Transferrin acquisition of catabolized erythrocyte iron in the rat

Rat plasma contains two isotransferrins rather than the single iron-binding protein found in plasma of other species, and it was recently proposd that differences between the biological behavior of each isotransferrin accounted for observations previously attributed to behavioral differences between each of the two transferrin iron-binding sites. The two isotransferrins were isolated from rat plasma by DEAE-Sephadex ion-exchange chromatography and isoelectric focusing. The pH-dependent iron-dissociating and reticulocyte iron-donating properties of each isotransferrin were investigated and found to be indistinguishable. Like human transferrin, one iron-binding site retains its affinity for iron below pH 6 and this property was used to investigate the $\text{in}$$\text{vivo}$ acquisition of catabolic iron in order to determine whether the process occurs at one specific or both binding sites. Plasma radioactive iron, derived from injected ⁵⁹Fe-labelled heat denatured erythrocytes was bound with high specificity to the transferrin iron-binding site that was most resistant to acidic dissociation. This finding supports Fletcher and Huehns' hypothesis that each of the two rat transferrin iron-binding sites is endowed with a separate functional role.     

Receptor-induced switch in site-site cooperativity during iron release by transferrin.

Iron removal by PPi from the N- and C-terminal binding sites of both free and receptor-complexed transferrin, when the partner site remains occupied with kinetically inert Co(III), has been studied at pH 7.4 and 5.6, at 25 degrees C. At extracellular pH, 7.4, the C-terminal site of free mixed-metal proteins is slightly more labile than its N-terminal counterpart in releasing iron to 0.05 M PPi. The rate and extent of iron removal are retarded from both sites when transferrins are receptor-bound. At endosomal pH, 5.6, the two sites exhibit greater kinetic heterogeneity in iron release to 0.005 M PPi. The N-terminal site is 6 times more facile in relinquishing iron than the C-terminal site when mixed-metal transferrins are free. However, the two sites are affected oppositely upon binding to the receptor. Iron release from the C-terminal site of receptor-complexed CoN-transferrin-FeC is 4 times faster than that from receptor-free protein. In contrast, iron removal from the N-terminal site of receptor-complexed FeN-transferrin-CoC is slowed by a factor of 2 compared to that from free protein. These results help explain our previous observation of a receptor-induced switch in site lability during iron removal from diferric transferrin at pH 5.6 (Bali & Aisen, 1991). Site-site cooperative interactions between the two sites of doubly-occupied transferrin during iron release are altered upon binding to receptor at pH 5.6. Iron in the otherwise weaker binding site of the N-terminal lobe is stabilized, while iron in the relatively stable binding site of the C-terminal lobe is labilized.     

Three-dimensional structure of mare diferric lactoferrin at 2.6 A resolution.

Lactoferrin is a monomeric glycoprotein with a molecular mass of approximately 80 kDa. The three-dimensional structure of mare diferric lactoferrin (mlf) has been determined at 2.6 A resolution. The protein crystallizes in the space group P 212121with a=85.2 A, b=99.5 A, c=103.1 A with a solvent content of 55 % (v/v). The structure was solved by the molecular replacement method using human diferric lactoferrin as the model. The structure has been refined using XPLOR to a final R -factor of 0.194 for all data in the 15-2.6 A resolution range. The amino acid sequence of mlf was determined using a cDNA method. The final refined model comprises 5281 protein atoms, 2 Fe3+, 2 CO32-and 112 water molecules. The overall folding of mlf is similar to that of other proteins of the transferrin family. The protein folds into two globular lobes, N and C. The lobes are further divided into two domains, N1 and N2, and C1 and C2. The iron-binding cleft is situated between the domains in each lobe. The N lobe appears to be well ordered and is more stable than the C lobe in mlf unlike in other lactoferrins, where the C lobe is the more stable. The opening of the binding cleft in the N lobe of mlf is narrower than those in other proteins of the transferrin family. This is very unusual and is found only in mare lactoferrin. Apart from certain hydrophobic interactions at the mouth of the cleft, one salt-bridge (Lys301 . . . . . . . . Glu216) crosses between the two walls of the cleft. The two lobes are connected covalently by a three-turn alpha-helix involving residues 334-344. The N lobe displays a highly ordered structure with appreciably low temperature factors. The iron coordination is more symmetrical in the N lobe than in the C lobe. There are only 16 intermolecular hydrogen bonds in the structure of mlf.     

Growth of human tumor cell lines in transferrin-free,low-iron medium

Summary Iron is essential for tumor cell growth. Previous studies have demonstrated that apart from transferrin-bound iron uptake, mammalian cells also possess a transport system capable of efficiently obtaining iron from small molecular weight iron chelates (Sturrock et al., 1990). In the present study, we have examined the ability of tumor cells to grow in the presence of low molecular weight iron chelates of citrate. In chemically defined serum-free medium, most human tumor cell lines required either transferrin (5 μg/ml) or a higher concentration of ferric citrate (500 μM) as an iron source. However, we have also found that from 13 human cell lines tested, 4 were capable of long-term growth in transferrin-free medium with a substantially lower concentration of ferric citrate (5 μM). When grown in medium containing transferrin, both regular and low-iron dependent cell lines use transferrin-bound iron. Growth of both cell types in transferrin medium was inhibited to a certain degree by monoclonal antibody 42/6, which specifically blocks the binding of transferrin to the transferrin receptor. On the contrary, growth of low-iron dependent cell lines in transferrin-free, low-iron medium (5 μM ferric citrate) could not be inhibited by monoclonal antibody 42/6. Furthermore, no autocrine production of transferrin was observed. Low-iron dependent cell lines still remain sensitive to iron depletion as the iron(III) chelator, desferrioxamine, inhibited their growth. We conclude that low-iron dependent tumor cells in transferrin-free, low-iron medium may employ a previously unknown mechanism for uptake of non-transferrin-bound iron that allows them to efficiently use low concentrations of ferric citrate as an iron source. The results are discussed in the context of alternative iron uptake mechanisms to the well-characterized receptor-mediated endocytosis process.     

The immune properties of Manduca sexta transferrin

Transferrins are secreted proteins that bind iron. The well-studied transferrins are mammalian serum transferrin, which is involved in iron transport, and mammalian lactoferrin, which functions as an immune protein. Lactoferrin and lactoferrin-derived peptides have bactericidal activity, and the iron-free form of lactoferrin has bacteriostatic activity due to its ability to sequester iron. Insect transferrin is similar in sequence to both serum transferrin and lactoferrin, and its functions are not well-characterized; however, many studies of insect transferrin indicate that it has some type of immune function. The goal of this study was to determine the specific immune functions of transferrin from Manduca sexta (tobacco hornworm). We verified that transferrin expression is upregulated in response to infection in M. sexta larvae and determined that the concentration of transferrin in hemolymph increases from 2 μM to 10 μM following an immune challenge. It is also present in molting fluid and prepupal midgut fluid, two extracellular fluids with immune capabilities. No immune-induced proteolytic cleavage of transferrin in hemolymph was observed; therefore, M. sexta transferrin does not appear to be a source of antimicrobial peptides. Unlike iron-saturated lactoferrin, iron-saturated transferrin had no detectable antibacterial activity. In contrast, 1 μM iron-free transferrin inhibited bacterial growth, and this inhibition was blocked by supplementing the culture medium with 1 μM iron. Our results suggest that M. sexta transferrin does not have bactericidal activity, but that it does have a bacteriostatic function that depends on its iron sequestering ability. This study supports the hypothesis that insect transferrin participates in an iron withholding strategy to protect insects from infectious bacteria.     

Isolation and molecular cloning of transferrin from the tobacco hornworm, Manduca sexta. Sequence similarity to the vertebrate transferrins

An iron-binding glycoprotein of Mr = 77,000 has been isolated from hemolymph of the adult sphinx moth Manduca sexta. Since this protein binds ferric ion both in vivo and in vitro and has a secondary structure similar to that of human serum transferrin and human lactoferrin as judged by CD spectra, we decided to clone its cDNA in order to determine its relationship to the vertebrate transferrins. Antiserum generated against this protein was used to screen a larval fat body cDNA library. A 2.0 kilobase clone was isolated that selects an mRNA which, when translated in vitro, produces an immunoprecipitable 77-kDa protein. When the library was rescreened using the 2.0-kilobase clone as a probe, three full-length clones were isolated, and the complete nucleotide sequence of one 2,183-base pair insert was determined. The deduced protein sequence contains an 18-amino acid signal sequence and a mature protein sequence of 663 amino acids with a calculated Mr of 73,436. The sequence was used to search the National Biomedical Research Foundation (NBRF) protein database, revealing significant similarity to the vertebrate transferrins, a family of 80-kDa glycoproteins which transport and sequester iron in the blood and other body fluids. A multiple sequence alignament shows the greatest areas of similarity to be around the two iron binding sites, although the insect protein seems to contain only one such functional site. Moreover, 23 of the 24 cysteine residues in the insect protein occupy identical positions as compared with the other transferrins, indicating a similar overall tertiary structure. Comparison of the two halves of the insect sequence indicates that the protein may have arisen as a result of gene duplication. The similarity of the M. sexta sequence to the vertebrate transferrins may provide important clues to transferrin evolution.     

Binding of phosphonate chelating agents and pyrophosphate to apotransferrin.

Difference ultraviolet spectroscopy has been used to monitor the binding of a series of phosphonate ligands to human apotransferrin. The ligands consist of pyrophosphate as well as the phosphonic acids (aminomethyl)phosphonic acid (AMPA), (hydroxymethyl)phosphonic acid (HMP), (phosphonomethyl)-iminodiacetic acid (PIDA), N,N-bis(phosphonomethyl)glycine (DPG), and nitrilotris(methylenephosphonic acid) (NTP). Equilibrium constants have been measured for the sequential binding of two ligands per molecule of apotransferrin. In addition, site-specific equilibrium constants have been measured for the binding of AMPA, HMP, and PIDA to the vacant binding site of both forms of monoferric transferrin. Since titrations of diferric transferrin produce no difference UV spectrum, it is proposed that the primary binding site for phosphonic acids includes the protein groups that bind the synergistic bicarbonate anion that is required for formation of a stable ferric transferrin complex. It is further proposed that those ligands with two phosphonate groups can simultaneously bind to cationic amino acid side chains that extend into the cleft between the two domains of each lobe of transferrin. From an inspection of the ferric transferrin crystal structure, the most likely anion binding residues in the cleft are Arg-632 and Lys-534 in the C-terminal lobe and Lys-206 and Lys-296 in the N-terminal lobe.     

Visual and olfactory input segregation in the mushroom body calyces in a basal neopteran, the American cockroach

The cockroach Periplaneta americana is an evolutionary basal neopteran insect, equipped with one of the largest and most elaborate mushroom bodies among insects. Using intracellular recording and staining in the protocerebrum, we discovered two new types of neurons that receive direct input from the optic lobe in addition to the neuron previously reported. These neurons have dendritic processes in the optic lobe, projection sites in the optic tracts, and send axonal terminals almost exclusively to the innermost layer of the MB calyces (input site of MB). Their responses were excitatory to visual but inhibitory to olfactory stimuli, and weak excitation occurred in response to mechanosensory stimuli to cerci. In contrast, interneurons with dendrites mainly in the antennal lobe projection sites send axon terminals to the middle to outer layers of the calyces. These were excited by various olfactory stimuli and mechanosensory stimuli to the antenna. These results suggest that there is general modality-specific terminal segregation in the MB calyces and that this is an early event in insect evolution. Possible postsynaptic and presynaptic elements of these neurons are discussed.     

Characterisation of recombinant unglycosylated human serum transferrin purified from Saccharomyces cerevisiae

Structural identity between a recombinant transferrin mutant (N413Q, N611Q) secreted from Saccharomyces cerevisiae and the native protein was shown by CD analysis and immunodiffusion assays against anti-hSTf. The ability of the recombinant protein to bind iron was confirmed by urea–PAGE and EPR analysis of the iron-saturated protein revealed the characteristic holo-transferrin spectrum, indicating conservation of both iron-binding sites. The integrity of the unglycosylated recombinant protein indicates that such protein could be a valuable tool not only for structure–function characterisation but also crystallisation assays. In addition, the recombinant transferrin was found to be as effective as native transferrin as a growth factor in cell culture medium.     

Structure of human lactoferrin: crystallographic structure analysis and refinement at 2.8 A resolution

The structure of human lactoferrin has been refined crystallographically at 2.8 A (1 A = 0.1 nm) resolution using restrained least squares methods. The starting model was derived from a 3.2 A map phased by multiple isomorphous replacement with solvent flattening. Rebuilding during refinement made extensive use of these experimental phases, in combination with phases calculated from the partial model. The present model, which includes 681 of the 691 amino acid residues, two Fe3+, and two CO3(2-), gives an R factor of 0.206 for 17,266 observed reflections between 10 and 2.8 A resolution, with a root-mean-square deviation from standard bond lengths of 0.03 A. As a result of the refinement, two single-residue insertions and one 13-residue deletion have been made in the amino acid sequence, and details of the secondary structure and tertiary interactions have been clarified. The two lobes of the molecule, representing the N-terminal and C-terminal halves, have very similar folding, with a root-mean-square deviation, after superposition, of 1.32 A for 285 out of 330 C alpha atoms; the only major differences being in surface loops. Each lobe is subdivided into two dissimilar alpha/beta domains, one based on a six-stranded mixed beta-sheet, the other on a five-stranded mixed beta-sheet, with the iron site in the interdomain cleft. The two iron sites appear identical at the present resolution. Each iron atom is coordinated to four protein ligands, 2 Tyr, 1 Asp, 1 His, and the specific Co3(2-), which appears to bind to iron in a bidentate mode. The anion occupies a pocket between the iron and two positively charged groups on the protein, an arginine side-chain and the N terminus of helix 5, and may serve to neutralize this positive charge prior to iron binding. A large internal cavity, beyond the Arg side-chain, may account for the binding of larger anions as substitutes for CO3(2-). Residues on the other side of the iron site, near the interdomain crossover strands could provide secondary anion binding sites, and may explain the greater acid-stability of iron binding by lactoferrin, compared with serum transferrin. Interdomain and interlobe interactions, the roles of charged side-chains, heavy-atom binding sites, and the construction of the metal site in relation to the binding of different metals are also discussed.